Letter Health Consultation ST. REGIS PAPER COMPANY SITE CASS LAKE, CASS COUNTY, MINNESOTA EPA FACILITY ID: MND057597940 APRIL 21, 2008 U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES Public Health Service Agency for Toxic Substances and Disease Registry Division of Health Assessment and Consultation Atlanta, Georgia 30333 Health Consultation: A Note of Explanation
An ATSDR health consultation is a verbal or written response from ATSDR to a specific
request for information about health risks related to a specific site, a chemical release, or the
presence of hazardous material. In order to prevent or mitigate exposures, a consultation may
lead to specific actions, such as restricting use of or replacing water supplies; intensifying
environmental sampling; restricting site access; or removing the contaminated material.
In addition, consultations may recommend additional public health actions, such as conducting
health surveillance activities to evaluate exposure or trends in adverse health outcomes;
conducting biological indicators of exposure studies to assess exposure; and providing health
education for health care providers and community members. This concludes the health
consultation process for this site, unless additional information is obtained by ATSDR which,
in the Agency’s opinion, indicates a need to revise or append the conclusions previously
issued.
You May Contact ATSDR TOLL FREE at 1-800-CDC-INFO or Visit our Home Page at: http://www.atsdr.cdc.gov LETTER HEALTH CONSULTATION ST. REGIS PAPER COMPANY SITE CASS LAKE, CASS COUNTY, MINNESOTA
EPA FACILITY ID: MND057597940 Prepared By: Minnesota Department of Health Under Cooperative Agreement with the U.S. Department of Health and Human Services Agency for Toxic Substances and Disease Registry Protecting, maintaining and improving the health of all Minnesotans
December 14, 2007
Mr. Timothy Drexler
Remedial Project Manager
Superfund Division
U. S. Environmental Protection Agency (EPA)
Region 5
77 West Jackson Boulevard (SR-51)
Chicago, IL 60604
Dear Mr. Drexler:
The Minnesota Department of Health (MDH) is submitting provisional comments on the
second Draft Human Health Risk Assessment for the St. Regis Paper Company Site, Cass
Lake, MN (HHRA) received from the International Paper Company (IP) on September 28,
2007. This second draft of the HHRA is far longer and complex than the first (2005) draft
(already long and complex). The additional material has little relevance as a response to
EPA’s specific required changes to the 2005 draft HHRA, and contains much added
material of little or no use. MDH did not review most of the calculations done by IP; but
it cannot be assumed that these were done correctly (as documented below). We may
have additional comments on this as well as other matters when we receive and have had
a chance to review comments from the Leech Lake Band of Ojibwe (LLBO), the
Minnesota Pollution Control Agency (MPCA) and the EPA. We are especially interested
in comments prepared by Eric Morton of TetraTech for EPA. These comments were
presented in a telephone conference call on November 19.
Our findings, spelled out in more detail in the review below, are that:
• The conclusions of the 2007 draft HHRA are not usable for decision-making;
• The risk assessment process has already been far too long; requiring IP to write
another draft is self-defeating because the third document is likely to be even
longer and more ungainly than the second document and past experience suggests
it will take another two years;
• EPA may (or may not) be able to use selected cancer and non-cancer risk
calculations from the current document for all exposure media analyzed including
fish as a starting point for its own calculations;
• EPA should include additional risks where appropriate (e.g., risk from surface
deposition onto garden crops, eating of fish eggs, exposures to shallow
groundwater, etc.);
St. Regis HHRA comments
MDH, December 14, 2007
Page 2
• EPA should appropriately consider uncertainties which suggest that the current
draft is biased and that risks are likely to be underestimated;
• EPA should use the risk calculations in conjunction with uncertainties and
generate a set of conservative remediation goals for soil and sediment;
• EPA should ignore IP’s specific risk calculations tied to specific exposure point
concentrations and should instead use its own set of validated environmental
sampling data to:
1) determine where more environmental sampling is needed in order to ensure
that the site is appropriately investigated in light of the remediation goals, and
then carry out the needed environmental sampling;
2) ensure that the site is cleaned up consistent with remediation goals; and
finally,
• Additional environmental sampling and subsequent remediation should be done
within a short time frame.
Specific Comments follow:
Chapter 1, Introduction:
Page 1-2. The conclusion that there are no site-related adverse risks for non-cancer
effects is obviously wrong (see below). The conclusion that the HHRA identified no siterelated cancer risks that exceed EPA’s acceptable risk range is also wrong (see below).
Page 1-2, page 1-14. It is implied that exposure scenarios evaluated in the HHRA for
both cancer and non-cancer effects include interim remedial actions to reduce exposures
to yard soil and house dust. But on page 1-14 the (presumably correct) assumption is
made that housecleaning and dust suppression applications will not continue once a
permanent remedy is selected. IP can’t have it both ways. In fact, it is inappropriate to
include interim remedial actions in any risk scenario, including current or future
scenarios. Modifying risk scenarios, especially exposure point concentrations, to take
into account temporary interim measures is inappropriate. Only the so-called “past”
scenario is valid for decision-making (but this is not even mentioned in the summary
statements).
Page 1-4. There is no mention that the site is located within the boundaries of a
reservation.
Page 1-12. Excavation and placement of contaminated material in containment vault: It
should be mentioned that large areas of the site were graded and soil was moved
throughout the site without proper site characterizations, resulting in dioxins and other
site contaminants being scattered across the site and deposited in high concentrations in
“remediated” areas. The use of uncharacterized fill material is problematic and has lead
to cross-contamination. The cross-contamination issues have been raised numerous times
by MDH.
St. Regis HHRA comments
MDH, December 14, 2007
Page 3
Page 1-15. The list of references used to conduct the HHRA should include the MPCA
PAH guidance for soil and air (http://www.pca.state.mn.us/air/aera-ci.html , and
http://www.pca.state.mn.us/publications/risk-tier2srv.xls ). The list should also include
MDH groundwater rule
(http://www.health.state.mn.us/divs/eh/groundwater/hrlgw/rules.html, and MN. Stat. Sec.
103H.201 - 103H.280) and MDH Health Risk Values for Air
(http://www.health.state.mn.us/divs/eh/risk/guidance/index.html). MDH has requested
numerous times that all soil, groundwater and sediment samples be analyzed for the full
set of carcinogenic PAH compounds found in the references listed above. It is troubling
that site operations utilized PAHs extensively, yet most of the samples were not analyzed
for the full list of carcinogenic PAHs needed to fully characterize human health risks.
Chapter 2, Data Evaluation and Selection
As with the first draft of the HHRA, the main problem with this risk assessment relates
back to the highly biased selection of data for evaluation.
A general comment concerns data use and selection criteria. There is a lack of
transparency in descriptions of sample validation, sample selection, statistical evaluation
and other steps in the EPC derivation. The reader is not able to follow each step and
duplicate the risk calculations in Chapter 4.
A general comment concerns the process for selecting COPC’s. The process for selecting
COPCs to be retained is biased to prevent a true assessment of cumulative exposures. If
a given chemical is detected in multiple media, but does not exceed the screening criteria
in any of those media it is completely ignored, even though the cumulative exposure may
be significant. Similarly, if a given chemical is detected above the screening criteria in
one medium, and is present but below the screening criteria in others, it is dropped from
consideration in those media, again masking the true level of exposures. If a chemical is
detected above the screening criteria in one medium, it should be included in the EPC
calculations using the concentrations in all of the media where it has been detected.
Similarly, if a chemical is detected in multiple media, but does not exceed the screening
criterion for any, then a “hazard index” calculation should be performed by dividing the
maximum concentration detected in a given medium by the screening criteria for that
medium and summing the values; if the sum is 1 or greater, the chemical should be
included in the EPC calculations using the concentrations in all of the media where it has
been detected. The only exception would be if the maximum concentration is within the
range of natural background concentrations detected in the reference areas, in which case
the contribution from that medium should not be included in the EPC calculations.
Fish data selection and risk assessment
These comments assume in general that the data and statistics are correct except where
noted, and rely on EPA’s evaluation of the completeness of the data and appropriate
St. Regis HHRA comments
MDH, December 14, 2007
Page 4
statistical treatment. The main focus was on dioxins in walleye and whitefish fillet and
the overall approach related to fish fillets.
Our overall conclusion is that risks from fish consumption should be included in the
total risk calculation. This conclusion is in agreement with comments of Eric Morton
of Tetratech in the telephone conference call of November 19, 2007.
Statistical analysis shows dioxins concentrations in walleye and tulibee/whitefish fillets
from the site are higher than reference sites. Higher concentrations in these species, in
particular, are of concern because they are the ones most consumed by tribal members.
Needing “confirmation” of differences is not appropriate for public health risk
assessment.
Differences in dioxin concentrations in fish between reference and site waters are
underestimated due to decisions about data use and data exclusions, including:
• Low frequency of detection for dioxin/furan congeners, e.g. page C-14 perchfillet data discussion. Site data had a 40% detection frequency and reference data
(LLBO Pilot Project) had only a 7% detection frequency.
• Small sample sizes.
• Exclusion of eggs and partials from exposure considerations. Page C-3 states
partials are not relevant for human health risk assessment. What evidence do they
have to support that statement? They say eggs were excluded due to small sample
size. IP was directed to use maximum detected values for small sample sizes (see
Data Issues Section of Change Items List from Tim Drexler). Also this is
inconsistent: other data sets with small sample sizes were included and evaluated.
Eggs were part of the agreed upon work plan and should therefore be included.
Eggs would not necessarily replace some fish meals, as implied (4-26); they could
be in addition to fish meals.
• Lack of reference data for whitefish; use of tulibee for reference comparisons.
• EPC for fish is calculated by combining all fish species data (p. 4-35, 4-37). The
report states that this is appropriate because no difference was seen between fish
species. Differences were examined between site and reference fish by species.
No evaluation of between species differences was reported. In addition it is well
known that contaminant concentrations are dependent on fish species. Data in C-1
indicates there are differences in TEQs between species in site and reference data.
• Split sample decision rules in general - averaging of split samples.
• Using SQL versus DL? If DLs are more similar between labs than SQLs, how
does use of ½ SQL rather than ½ DL to censor non-detects for dioxins impact
assessment of differences between reference and site fish?
St. Regis HHRA comments
MDH, December 14, 2007
Page 5
• Excluding samples where detection limit exceeded highest detected value.
• Analysis of congener profiles may reveal further differences between site versus
reference fish. Due to the timeframe for this review this analysis was not carried
out.
• Excluding reference lakes from the NLFTS due to trophic status. There are
additional walleye data from lakes that could logically be included as reference.
Including these four lakes reduces the mean TEQ concentration for walleye from
the NLFTS. The total TEQs (ND=0) for all four excluded lakes were lower than
the any of the included lakes. Cass Lake had the highest total TEQ (ND=0). The
TEQ DF (ND=0) from Cass Lake is an order of magnitude higher than walleye
from the other NLFTS sites.
NE Minnesota Walleye Data from NLFTS
NLFTS
SampleID
MN020980PS
MN000460PS
MN021110PS
MN011010PS
MN990081PS
MN000180PD
MN030933PS
MN000180PS
MN030235PS
MN990205PS
Site_Name
Red Lake
South
Mcdougal Lake
Vermilion Lake
White Iron Lake
Fox Lake
Woman Lake
Mille Lacs
Woman Lake
Snowbank Lake
Cass Lake
Total
TEQ
ND=0
0.006293
TEQ DF
ND=0
0
0.011022
0.015917
0.018162
0.018526
0.038292
0.040019
0.048318
0.058761
0.098134
0
0.003
0
0
0.02
0
0.003
0
0.037
* first four lakes were not included as reference sites in risk assessment
Selection of Reference Lakes from the NLFTS
MDH comments from January 30, 2006 on the selection of reference lakes were not
addressed. It is still not clear if NLFTS data for Cass Lake walleye were included in the
risk assessment; this data is not listed in Table C-1. In general, the basis for the selection
criteria is not adequate. Excluding a lake based on trophic status is not justified. In fact
dioxin TEQ data from the NLFTS do not show a relationship between dioxin TEQ and
trophic status.
St. Regis HHRA comments
MDH, December 14, 2007
Page 6
Oneway Analysis of TTEQNDZERO By TS Composite_Type=Predator
0.3
TTEQNDZERO
0.25
0.2
0.15
0.1
0.05
0
E
M
M/E O
TS
Missing Row s
18
Quantiles
Level
E
M
M/E
O
Minimum
0.002256
0.012319
0.160664
0.029976
10%
0.005082
0.013019
0.160664
0.029976
25%
0.011597
0.020347
0.160664
0.035363
Median
0.026577
0.035656
0.160664
0.044649
75%
0.109517
0.048477
0.160664
0.098271
90%
0.227632
0.066497
0.160664
0.098682
Maximum
0.259712
0.110766
0.160664
0.098682
Inconsistencies noted in HHRA Report:
Table 2-7
All data sets considered in this analysis are presented in Tables 2-1 through 2-7 (fish is 2­
7) (page 2-3 section 2.2 last sentence). EPA NLFTS samples are not listed in Table 2-7.
Egg samples are listed in Table 2-7 and some of these have Dioxins/Furans marked with
an “X” indicating to include in risk assessment.
Tables 2-18 and 2-19
Table 2-19 lists reference samples to exclude but doesn’t appear to be consistent with
Table 2-7. For example for Sample Event 2002 LLB2, Table 2-7 has an “X” under
Dioxins/Furans for samples 02LLW009 and 02LLW010 while Table 2-19 lists these
same samples as excluded.
Table C-11. Results from Wilcoxon Comparison for Site vs. Reference.
Project specific reference walleye n=3 so how can table C-11 say “site is not greater”?
According to text on C-9 statistical tests were performed only when both data sets
contained at least five values.
St. Regis HHRA comments
MDH, December 14, 2007
Page 7
Soil Data Selection and Risk Assessment
Considering the site was a wood treatment facility that used PAHs extensively, why were
very few soil samples and other media analyzed for the full list of carcinogenic PAHs?
Why have so few soil samples been collected deeper than 12 inches on such a large site?
These comments have been made numerous times dating back to the EPA investigation
in 2001. Additionally, the MPCA’s Risk Based Evaluation Guidance Manual defines the
first 4 feet as accessible, and the 4-12 feet (below ground surface) as potentially
accessible soil.
IP excluded soil data from consideration for reasons having nothing to do with “rules”
described in Chapter 2. Table 2-1 gives numerous examples with comments such as the
following:
“Excluded: Excavated in 2004 removal action.” What are the soil concentrations at the
base of the excavations? The removal action removed soil to a depth of 12 inches or less
in most areas. These data should be included in the risk calculations.
“Sample of stained material in roadway collected at unknown depth.” These data should
be used, and it should be assumed that they were surface samples.
“Excluded: Topsoil used to backfill excavated areas on the Allen property during the
removal.” These data should be included. Many excavations were to a depth of as little
as 4 inches.
“Split selected: Positive bias in split sample results.” EPA should verify that the results
with the higher value should be selected.
“Covered by a geotextile fabric and a minimum of 4-inches of gravel per the work
plan…” This is an inadequate excavation depth. The sample should be included.
“Excluded: property burned down.” This is not a reason to exclude a sample.
Sediment Data Selection and Risk Assessment
The manner in which reference area samples are compared to samples at/near the site is
not valid and is used to disqualify data that should be included in the EPCs. Barr/IP
apparently compared the entire “site area” data set to the entire “reference area” data set.
But the “site area” data set contains results from areas both near and far removed from
the source areas, creating a much larger range of sample results that includes background
levels. Comparison of reference data to particular site areas, especially those closer to
sources, is far more revealing. For example, when the reference area arsenic sediment
data set (ND – 13.1 mg/kg) is compared to the overall site area sediment data set (ND –
29 mg/kg), there is no “statistical” difference between them. But when the reference area
St. Regis HHRA comments
MDH, December 14, 2007
Page 8
data set is compared to the sediment data from Fox Creek near the city dump (9 – 29
mg/kg), the difference is obvious. Clearly, the sediment in Fox Creek near the city dump
has higher arsenic levels than the background levels in the reference area.
Table D1-11.
No sediment samples have been analyzed using the extended PAH list – doesn’t this have
to occur to determine if these compounds need to be included in the COPC?
Groundwater Data Selection and Risk Assessment
In 1998, the list of PAHs for which groundwater samples were analyzed was significantly
shortened. As noted in the MDH draft Public Health Assessment for Groundwater,
Surface Water and Sediments (December 2005), the PAHs dropped from the analyte list
accounted for a significant percentage (24-90%) of the nPAH and Total PAH in 24
monitoring wells. While most of these compounds did not exceed human health criteria,
two did: carbazole (wells 118, 402, 405, 409, 2401, and 2403) and dibenzofuran (wells
118, 207, 401, 402, 405, 409, 2401, and 2403). These compounds should be retained as
COPCs and included in the EPC calculations using the concentrations in any media
where they were detected.
Furthermore, groundwater PAH concentrations should be compared to the MDH Total
Petroleum Hydrocarbon Health Based Value.
Surface Water Data Selection and Risk Assessment
Table D1-12.
PCP was excluded from the COPC list, even though it was detected in 40% of the
samples. PCP is one of the primary COPCs for the site, and exposure to it occurs through
multiple pathways; it should be retained as a COPC and included in EPC calculations in
order to evaluate cumulative exposures.
Chapter 3, Chemical Sources, Release Mechanisms, and Transport Pathways
3.1.2.3, p. 3-8. Second and third paragraphs reference dumping of sludge in the
southwest area, but Figure 1-2 does not show any dumping locations mentioned in the
text.
Chapter 4, Human Health Risk Assessment
General Comments
• The HHRA should not include the interim remedial action (4 inch soil cover)
as a dilution factor. On October 28, 2003 Patricia Bloomgren, Director of the
St. Regis HHRA comments
MDH, December 14, 2007
Page 9
Division of Environmental Health, wrote to William Muno, Director of the
EPA Superfund Division: “The removal action should be taken so that it
complements, rather than interferes with, any future remedial actions
performed as a result of human health and ecological risk assessments.” The
letter went on to state: “Soil removal at this depth [4 inches] will not protect
gardeners, children digging, people playing sports, or doing any number of
ordinary activities. It will not prevent wind erosion from uncovering more
contaminated soil.” On June 23, 2005, Rita Messing, Supervisor of Site
Assessment and Consultation and Daniel Peña, Health Assessor in the same
unit, wrote to Tim Drexler, Remedial Project Manager: “…MDH believes
that it is important to carry out the interim plan in such a way that the ultimate
remedial action is not compromised. The interim plan calls for covering and
seeding residential yards. MDH recommends that soil levels used to evaluate
remedial actions not include the dilution caused by the interim action….” The
properties included in the IRA should not be referred to as remediated in the
text.
• It is unacceptable for the residential exposure scenarios to exclude fish
consumption.
• It is prudent and reasonable to consider the construction worker and utility
worker to be a resident. A site resident is currently a public works employee
of Cass Lake. Additionally, in a small town, the construction worker and the
utility worker are likely the same person – in Cass Lake they may also be a
tribal member.
• The St. Regis Superfund site HHRA should use the same garden exposure
factors used in EPA’s HHRA for the South Minneapolis Neighborhood Soil
Contamination Superfund site. Treating the two populations differently in
terms of garden vegetable consumption is not justifiable.
Conceptual Site Model
4.1.1, p. 4-1. The entire site area is a reservation. This was omitted from the
description.
4.1.3, p. 4-4. The section does not adequately discuss or present any reasonable
institutional controls as required by MN Uniform Environmental Covenants Act, Minn.
Stat. Ch. 114E (Supp. 2007), and other MN rules and regulations. Any institutional
controls will have to be coordinated with both the City of Cass Lake and the Leech Lake
Band of Ojibwe. Until that time, the future land use assumptions for the entire site remain
questionable at this time. In fact, the City of Cass Lake zoning ordinance permits public
parks and playgrounds, as well as any number of commercial uses in the area described
as being reserved for industrial use.
St. Regis HHRA comments
MDH, December 14, 2007
Page 10
4.1.3.1, p. 4-5. The assumption that subsurface soils are generally lower in concentrations
than surface concentrations is highly uncertain due to cross-contamination issues and
very limited number of soil samples collected deeper than 12 inches. Based on the
summary review of the soil date evaluated for use in the HHRA, it appears that only
three soil samples deeper than 12 inches were evaluated in the HHRA (see table 2-1).
Furthermore, it appears that only one (J29-29 (12 -24)) of the three samples was analyzed
for PAHs, and all were located in the treatment facility operable unit. The lack of soil
data deeper than 12 inches below grade has been raised by MDH numerous times dating
back to the EPA site investigation in 2001.
The current and future residential exposure scenario must include ingestion of both
surface and root crops (see general comments). Dioxin root uptake by garden plants is not
the only pathway through which residents can be exposed. Surficial deposition of dioxin
contaminated dust on the exterior of edible garden greens and plants in the tribal lifeways
and residential scenarios must be included in risk calculations.
4.1.3.1, p. 4-6. The residential exposure scenario does not include dermal contact with
contaminated groundwater. Groundwater is very shallow over much of the eastern
portion of the site, even expressing itself as surface water in some of the wooded areas.
We know that children have dug holes as deep as two feet in some portions of Area A,
which could allow them to come into contact with the groundwater. Dermal contact with
groundwater should be part of the residential exposure scenario, at least for children, if
not adults.
4.1.3.4, p. 4-7. Under the recreational scenario the report says people “might” be using
non-residential properties in Area A for recreation. There is so much obvious evidence
of this usage, that the report should not qualify this and instead say that people “are”
using those areas.
4.1.3.3, p. 4-7. It is reasonable to assume that the current and future worker/utility worker
receptor could also be a site resident and or a tribal member. A current site resident is
also a public works employee for the City of Cass Lake.
4.1.4, p. 4-8, and 4.1.4.2, p. 4-10. It is reasonable to assume that the current and future
worker/utility worker receptor could also be a site resident and/or a tribal member.
Currently, a site resident is also a City of Cass Lake public works employee.
4.1.4.1, p. 4-9. Was contaminated dust inclusion in honey considered in the
subsistence/tribal lifeways exposure scenario?
Identification of Chemicals of Potential Concern
4.2.3.1, p. 4-13. The last paragraph incorrectly lists the MPCA Residential Soil Reference
Value for Dioxin as 200 ppt. The referenced St. Regis Soil Health Consultation lists the
dioxin SRV as 20 ppt in Appendix F (see
St. Regis HHRA comments
MDH, December 14, 2007
Page 11
http://www.health.state.mn.us/divs/eh/hazardous/sites/cass/stregis/soilshcapendf_h.pdf ).
The current dioxin SRVs are 20 ppt and 35 ppt for residential and industrial settings,
respectively (see http://www.pca.state.mn.us/publications/risk-tier2srv.xls). Please list
the correct MPCA Dioxin SRVs throughout the HHRA.
4.2.3.2, p. 4-14. Barr has created their own sediment screening criteria, based on soil
PRGs and adjusted for lower exposure frequency, that are orders of magnitude larger than
the values developed by MDH for the U.S. Steel site and other PAH sites. They assume
shorter skin contact than with dry soil, when in fact sediment is likely to adhere longer
than dry soil because it is wet. They apparently do not include any factor for incidental
ingestion or inhalation as part of their “screening criteria”, even though these pathways
will be completed for several of the exposure scenarios where sediments are encountered
(recreational, subsistence or traditional tribal life-ways).
4.2.4.1, p. 4-16-17. The second paragraph does not note the residential area north of the
tracks as exceeding the 50 ppt dioxin screening level. Additionally, the referenced
Appendix Table D1-1 does not note the correct Region 9 EPA Residential Dioxin PRG.
Table D1-1 through D1-6 do not list all the carcinogenic PAHs used by the MN to
evaluate cancer health risks from exposure to wood treatment facilities. The tables do not
clearly describe what carcinogenic PAHs were evaluated for each exposure scenario. The
same can be said for Table 4-1. Without this critical information the reader is not able to
evaluate if the cancer risk has been fully characterized.
See also comments above for Groundwater, Surface Water and Sediment Data Selection.
4.2.4.2 and Table D1-11. PCP was excluded from the COPC list, even though it was
detected in 16% of the samples. PCP is one of the primary COPCs for the site, and
exposure to it occurs through multiple pathways; it should be retained as a COPC and
included in EPC calculations in order to evaluate cumulative exposures.
4.2.4.2, p. 4-18. As described above in the comment on selection of sediment data, the
comparison to reference area sediment samples to site samples has been used
inappropriately to disqualify arsenic from inclusion as a COPC.
Section 4.2.4.2, p. 4-18: The report incorrectly suggests that only samples from Fox
Creek and the channel exceed their screening levels for dioxins/furans. Using the
screening standards developed by MDH for use at the U.S. Steel site, the Fox Creek delta
at Pike Bay, the wetlands adjacent to the channel, deep hole #1 in Pike Bay, and deep
hole #1 in Cass Lake (EPA samples, 2001) all exceeded the screening levels by 3-5
orders of magnitude, and exceeded the reference area concentrations by 1-3 orders of
magnitude. All of these areas need to be considered in the EPC calculations.
St. Regis HHRA comments
MDH, December 14, 2007
Page 12
4.2.4.4, p. 4-19. The third paragraph seems to suggest that consideration of dioxins in
groundwater for the utility worker scenario should be ignored because it was likely
associated with NAPL in the wells where detected. If utility workers encounter areas of
the site where NAPL is present, how will the associated levels be accounted for if not
through the groundwater component of that scenario? And, as discussed above
regarding Area A exposure scenarios, children have been known to dig deep holes while
playing in non-residential portions of Area A. Exposure to dioxins via groundwater
dermal contact should be included in the residential EPC calculations for children.
4.2.4.4, p. 4-19. The HHRA has not considered the use of MN’s Health Based Value for
total petroleum hydrocarbons (200 ug/l, pyrene surrogate) as a screening value. The
tribal lifeways scenario should consider groundwater as a source of drinking water.
4.2.4.4, p. 4-19 - 4-20. The presence of NAPL in W2401, and detections of individual
PAHs above their solubility limits, may be a reason for excluding the 2006 data from this
well. However, it should not be used to further exclude all detections in that well of
individual PAHs that also have been detected elsewhere in the groundwater (i.e.
acenaphthene, fluoranthene, and pyrene), particularly as these compounds have similar
target organs (acenaphthene and fluoranthene – GI Tract/liver; fluroanthene and pyrene –
kidney) and detections even below the HRL or MCL could result in additive
concentrations of concern.
Exposure Assessment
4.3.1.1, p. 4-23. As discussed above, children may have dermal contact with
groundwater in the non-residential portions of Area A. This should be added to the
residential scenario.
4.3.1.3, p. 4-24. It is reasonable for the onsite worker and utility worker to be the same
person, and they may also be a site resident.
4.3.2.1, p. 4-26. Eggs would not necessarily replace some fish meals, as implied; they
could be in addition to fish meals.
4.3.2.1, p. 4-29. It is not practical to use BaP comparison concentrations that cannot be
verified as wet or dry weight.
4.3.3.1, p. 4-34. In general MDH agrees with the last sentence in the 3rd paragraph:
“Concentrations from the individual or multiple residential properties were not
aggregated to calculate an area-wide average because people are expected to have the
majority of their soil contact on their own properties.” This should also apply to the
residential samples collected north of the railroad tracks. One of the composite sample
results, composed of several residential properties, exceeds the ATSDR dioxin 50 ppt
screening level, and the MPCA dioxin SRV. Therefore, MDH requests that individual
St. Regis HHRA comments
MDH, December 14, 2007
Page 13
yards north of the tracks be sampled for dioxin, so each resident can be given yard
specific exposure advice.
4.3.3.2, p. 4-36. The use of the interim remedial action (3-4 in) soil cover as a dilution
factor is not acceptable (see general comments). The MPCA considers the first 4 ft soil
accessible in a residential setting. It is not clear that the soil samples were analyzed for
full list of carcinogenic PAHs. The Barr 2006 table 3 reference does not match the text
discussion of soil TEQdf and BaPEs results. The Barr 2006 reference is a groundwater
monitoring report. Second paragraph. The Allen property clearly exemplifies how dioxin
concentrations can vary wildly due to cross contamination and soil alterations. Why is
the Allen property, potentially the most dioxin contaminated residential property, still
lacking BaPE results?
4.3.3.2, p. 4-35, 4-37. EPC for fish is calculated by combining all fish species data. The
report states that this is appropriate because no difference was seen between fish species.
Differences were examined between site and reference fish by species. No evaluation of
between species differences was reported. In addition it is well known that contaminant
concentrations are dependent on fish species. Data in C-1 indicates there are differences
in TEQs between species in site and reference data.
4.3.3.2, p. 4-36 to 4-37. The averaging of the soil data from the Allen residence (as
detailed in the footnote on p. 4-37) serves simply to further reduce the EPC. The results
are already derived from averaged composite samples, and a second averaging is not
acceptable. EPC’s should not be “recalculated.” Original, pre-IRM data should be used
throughout.
Section 4.3.3.2, p. 4-37. Combining data from Fox Creek, the channel, and Pike Bay
may be helpful from a statistical perspective, but is not helpful from a public health
perspective. People may use all three areas, or they may use only one – if that one area is
Fox Creek, they will have a greater exposure than if they “average” their use across the
three areas. An EPC for use of all the areas, but in order to develop an RME, an EPC
should also be calculated for the worst of these areas, and that would be Fox Creek near
the city dump.
4.3.3.2, p. 4-39. In order to distinguish site from background, especially for evaluation of
fish data, in order to remove “noise,” TEQ’s should be calculated using zero to
characterize non-detected congeners. This comment has been made repeatedly over the
past 5 years. See also comments above for fish data evaluation.
4.3.3.2, p. 4-39. Last paragraph. The lack of deeper soil samples is an unacceptable data
gap dating from the EPA 2001 investigation and perpetuated with the IP investigation. It
is important for a soil investigation to have sufficient data to characterize extent and
magnitude of contaminants both vertically and horizontally in order to carefully define
risks to human health.
St. Regis HHRA comments
MDH, December 14, 2007
Page 14
4.3.3.2, p. 4-40. It is inappropriate for IP to state that “subsurface soil concentrations are
typically lower than surface soil concentrations at the same location” when only three soil
samples deeper than 12 inches were evaluated in the HHRA. More troubling still is IP’s
conclusion that “restricting the HHRA to surface soil concentrations likely results in
conservative (protective) risk estimates.” IP’s conclusions based on the soil data are
highly uncertain and do not take into account cross contamination issues. Table 4-2.
Summary of Uncertainties states that “the exclusion of subsurface soil data has a low
potential to overestimate risk.” IP asserts that “surface soil concentrations tend to be
higher than subsurface concentrations. Activities that mix surface and subsurface soil
would tend to decrease concentrations remaining at the surface.” This conclusion is not
substantiated when subsurface soil data are virtually nonexistent. IP was asked to change
this conclusion in the first draft.
4.3.3.3. p. 4-41. IP utilizes a vegetative cover value of 0.5. The MPCA SRV calculations
use the following vegetative cover values:
• Residential scenario = 0.5
• Industrial scenario = 0.0
• Recreational scenario = 0.25
These values are specially suited for the St. Regis site where there are sandy soils, dirt
roads, and most of the site is uncovered.
Section 4.3.3.3, p. 4-44. Is it reasonable to assume that utility trenches will be only 4 ft.
deep simply because that is the depth to groundwater? Given the depth of frost in
northern Minnesota, most utilities are buried 6 feet deep and presumably they could use
pumping to allow excavation below the water table.
4.3.3.3, p. 4- 46 to 4-47. It is not acceptable that IP excluded the consumption of surface
grown vegetables and greens. The ingestion of surface garden vegetables is part of RME
residential exposure scenario. It is misleading to present the residential exposure scenario
as including garden vegetable ingestion when it does not include surface grown crops.
4.3.4.1, p. 4-52. It is reasonable to assume that the utility worker (short-term exposure)
and the construction worker (long-term exposure) might be the same person in a small
town, and also a site resident.
4.3.4.2, p. 4-53. The exposure frequency of 83 days/yr is not protective enough for a
resident. The HHRA should not use a recreational exposure duration to represent a
residential scenario. A resident is likely to be outside in their yard most everyday of the
year when the weather is reasonable. Residents spend more time in their yards than time
spent away recreating.
St. Regis HHRA comments
MDH, December 14, 2007
Page 15
4.3.5.1, p. 4-60. The referenced EPA document (Guidance Manual for the Integrated
Exposure Uptake Bio-kinetic Model for Lead in Children, 1994) for partitioning dust and
soil exposure states the following:
“.. the option to allocate a portion of the ingested dust to dust derived from
soil that is ingested during outdoor play activity … is important when
there are differences between the bioavailability of dust derived from soil
and dust in the home, when house dust is thought to be mostly of soil
origin and each are expected to have similar biovailability, the designation
of this fraction is a moot point. It is in cases where house dust differs
significantly from soil derived dust that soil/dust ratio becomes
important.”
Indoor dust at the site is derived from out door soil. Additionally, smaller dust particle
size and increased digestible organic content found in dust generally result in increased
bioavailability. IP should therefore increase the bioavailability factor used in the house
dust dose calculations. The partitioning between dust and soil is an unnecessary
complication.
4.3.5.1, p. 4-63. The USEPA (2002h) reference defining a construction worker and
outdoor worker does not take into consideration the likelihood that workers in small
towns can be site residents who perform the tasks of both the construction and outdoor
workers.
4.3.5.2, p. 4-64. IP appears to use a soil/dust dioxin relative bioavailability factor (RBA)
of 0.5. The PCA uses a 0.55 RBA for dioxin in soil. It is also reasonable to utilize a larger
RBA for dusts. Other RBA differences noted are as follows:
Contaminant
Napthalene
BAP
PCP
IP Relative
Bioavailability
0.6
0.6
0.9
MPCA Relative
Bioavailability
0.8
0.8
1.0
MDH recommends that using MPCA Soil Reference Value RBAs in the HHRA. Also
note that text RBA values do not match the table 4-11 values.
Parameter Value Tables, Pp. 4-67, 4-76, 4-79, 4-90, 4-96, 4-99, and 4-101. Please list all
parameter values for each scenario in the tables for clarity. Referring the reader to the
text to find parameter values is burdensome and not helpful.
4.3.6.3, p. 4-72. The following table summarizes IP’s soil adherence parameter values
that differ from the MPCA SRV soil adherence factors (RME) :
Receptor
IP AF
MPCA AF
St. Regis HHRA comments
MDH, December 14, 2007
Page 16
Child (< 6 – 18 yrs)
Adult (>18-33 years)
Adult
Child (< 6 – 18 yrs)
Residential Scenario RME
0.2
0.07
Recreational Scenario RME
0.07
0.2
0.2
0.13
0.35
0.35
4.3.6.3, p. 4-73. The suggestion that sediments underwater are “likely to be washed off
the skin before the individual reaches shore” is contrary to the experience of anyone who
has ever waded or swam in an area with silty or muddy sediments and returned to shore
with silt and clay clinging to their legs and feet.
4.3.6.3, p. 4-74. Why is it “unlikely that loadings of Fox Creek and the channel area
sediment on skin would reach the monolayer coverage required to attain the adherence
factors found in the Shoaf et al. studies”?
4.3.11, p. 4-99. Why does IP’s HHRA not include consumption of surface grown produce
such as salad greens, tomatoes, and herbs? Region 5 EPA’s HHRA for residents living on
the South Minneapolis Superfund site includes the consumption of both root and surface
grown vegetables. Cass Lake residents should not be treated differently than South
Minneapolis residents in terms produce consumption rates, exposure frequency, and
exposure duration parameters without justification.
Toxicicity Assessment
The toxicity assessment for dioxins is not in agreement with State of Minnesota policy.
This policy appears at:
http://www.health.state.mn.us/divs/eh/risk/guidance/dioxinmemo2.html , and
http://www.health.state.mn.us/divs/eh/risk/guidance/dioxinmemo1.html).
These policies are also attached to this document.
IP also refers several times to the NAS/NRC report (2006). We are attaching slides from
a talk given on September 17, 2007 by David Eaton, the chair of the NAS committee, at a
conference at Michigan State University. These slides summarize salient points from the
report. We are also attaching slides from a talk given by Michael De Vito, Chief,
Pharmacokinetics Branch, EPA at the same conference. Conference proceedings are
available at:
http://cit.msu.edu/Superfund/Dioxin%20Workshop.html
More specific comments follow.
St. Regis HHRA comments
MDH, December 14, 2007
Page 17
4.4.2 and 4.4.2.1, p. 4-107-108. We agree with the statement that non-cancer endpoints
may be a greater public health issue. However, ATSDR’s chronic MRL (which is used in
the HHRA, and is based on reproductive effects in the Rhesus monkey) may not be
adequately conservative. It is simply the best evaluation that we have at the present time.
DeVito (slides 20,21) indicates that other sensitive endpoints could be chosen, and that
perhaps the immune system may be the most sensitive endpoint. Furthermore, human
body burdens of dioxins are already at or near the ED01 for many of these non-cancer
effects, so that the “margin of exposure” for non-cancer effects of dioxins is small and
possibly even negative. These calculations are not based on the traditional RfD divided
by uncertainty factor approach but are based on calculating equivalent body burdens for
dioxins. This is the approach recommended by NAS (Eaton, slide 34).
4.4.2.2, p. 4-108-109. IP neglects to mention that just as the NAS/NRC did not endorse
the EPA revised cancer slope factors (1E+6/(mg/kg-d) based on human data and 1.4
E+6/(mg/kg-d) based on animal data, neither did the NAS/NRC endorse the old slope
factor of 1.5/(mg/kg-d). The point is that EPA does not have a cancer risk assessment
policy. The State of Minnesota, however, does have such a policy. It is to use the slope
factor based on animal data: 1.4E+6/(mg/kg-d). The reasons for this are spelled out in
the attached policy document. MPCA has used this policy document to calculate Soil
Reference Values for dioxins in soil of 20 ppt (residential) and 35 ppt (industrial). For
present purposes, it is sufficient to note that this slope factor is neither the lowest nor the
highest factor that could be derived (see for example Eaton’s slides. Most importantly, it
incorporates use of body burden as a dose metric: an approach endorsed by the NAS
committee (cf. Eaton’s slides). The old slope factor does not incorporate this more
modern approach, and hence is not regarded as a valid decision-making tool in
Minnesota.
Thus, as a minimum, MDH concludes that the higher of the two slope factors used in the
HHRA should be chosen for decision-making: i.e., 1E+6/(mg/kg-d). The State of
Minnesota may choose to re-calculate cancer risks using the preferred slope factor of
1.4E+6/(mg/kg-d) as we develop recommendations for further sampling and site
remediation.
4.4.2.3, p. 4-110-111. From the standpoint of risk assessment it does not matter whether
2,3,7,8-TCDD is classified as a human carcinogen or as a probable human carcinogen.
As Dr. Eaton says in his slides, “Committee felt that it really was not important, as
TCDD will (and should) be regulated as if it is carcinogenic to humans regardless of what
label it is given.” Furthermore, Dr. Eaton says: “Overall, the committee concurs with
the value of conducting analyses of total cancers given the potential for dioxin to affect
multiple types of cancer.” Thus, the attempt in the HHRA to cast doubt on this finding
(e.g., “lack of consistency among specific tumor studies,” should be ignored.
4.4.2.4, p.4-111-113. A non-linear model for assessment of carcinogenicity of dioxins is
not available. IP seems to make the assumption that such a model would result in a
lowering of calculated cancer risks. This is not yet determined. In view of the existing
St. Regis HHRA comments
MDH, December 14, 2007
Page 18
body burden in the population, it is entirely possible that a non-linear model could lead to
higher calculated risks. This section is irrelevant for current decision-making.
4.4.2.5, p. 4-113-114. IP reviews the criticisms of EPA’s decision to use an ED01 as a
point of departure (POD) for calculating cancer potency. IP also concurs with the NAS
suggestion that EPA review the NTP animal bioassay (not available when EPA did its
reassessment). Again, it is unclear how changing the POD or reviewing the NTP study
will affect the determination of a potency slope, so this is irrelevant for current decisionmaking.
4.4.2.6, p. 4-115-117. IP reviews recent papers suggesting that the human-derived cancer
potency estimate of 1E+6 is too high because half-life assumptions used to back calculate
original exposures are too long. However, the papers cited need critical review by EPA
and public health agencies before their conclusions can be accepted by regulatory and
other government agencies entrusted with protecting public health. The DeVito slides
suggest that this is underway, and that the issues are more complex than depicted in the
HHRA. The section is irrelevant for current decision-making.
Risk Characterization
The risk characterization is not useful for regulatory decision-making because the preinterim action conditions are not considered. It is not useful because fish consumption is
not included in the exposure scenarios. It is not useful because of the exclusion of
important data.
4.5, p. 4-126. The conclusion that concentrations of contaminants in fish are not higher
than concentrations of these chemicals in background reference lakes is wrong.
4.5.1.1., p. 4-128. It is unjustified to assume that mixing of the top foot yard soil results
in lower surface dioxin concentrations. It is very troubling that IP has dismissed germane
data in the HHRA. For example soil sample EPA 2001 RES-16 was excluded “because
the property burned down” (see table 2-1). Why are samples RES-16, RES-16A, and
RES-16B missing from the hazard index risk summary table for past pre-IRM conditions
table (see table D4-7a)? There were residents living on the property in the past and RES­
16 is representative of the pre-IRM conditions. How can sample RES-16A pose the
maximum noncancer hazard if the other samples appear to have higher contaminant
concentrations (see table below)?
Sample
TEQdf (ppt) nd=0 PAH data PCP (ppb)
RES-16 (EPA, 2001)
485
yes
1900
RES-16A (IP, 2003)
287
No data
No Data
RES-16B (IP, 2003)
48
No data
No Data
St. Regis HHRA comments
MDH, December 14, 2007
Page 19
4.5.1.1, p. 4-128. “Consumption of homegrown produce was not considered for current
conditions because there are no gardens grown in unamended soil.” Again, the workplan
requires that homegrown produce be considered in the risk assessment. Again,
assessment under “current conditions” is deficient. A remedial action is done to protect
public health and the environment for all relevant future uses. Houses burning down,
inadequate 4 inch ground covers, etc. are irrelevant.
4.5.1.2. , p. 4-130. The following statement illustrates the degree of data manipulation
and opacity of hazard and risk calculations in the HHRA: “The higher the pre-IRM soil
concentrations, the greater the degree of hazard reduction achieved by the IRM. For both
children and adults, the current hazard index at the location with minimum pre-IRM soil
concentrations (RES-05) is 30% of the pre-IRM value (Table 4-16). The current hazard
index at the location with the maximum pre-IRM soil concentration that also underwent
interior house cleaning (RES-09) is 0.4 percent of the pre-IRM value.” Are we therefore
to assume that hazard indices are underestimated by factors between 3.3 and 250? These
numbers highlight the lack of transparency in the risk calculations: based on the dilution
factors described in the HHRA one would not expect these pre-post differences.
MDH suggests that EPA disregard IP’s calculations of hazard indices from exposure
point concentrations. EPA should utilize the original, pre-interim action data, and
calculate soil concentration that yields a hazard index of 0.05, 0.1, 0.2 and 1.0.
4.5.1.3, p. 4-131. Conclusions for Area B non-cancer hazards do not reflect consumption
of fish. They do not adequately reflect hazards to an individual spending most of their
time in the Fox Creek area.
4.5.1.4, p. 4-132-133. Conclusions for Area A/B combined do not reflect fish
consumption or pre-interim action conditions. They do not adequately reflect hazards to
an individual spending most of their time in the Fox Creek area.
4.5.2.1, p. 4-135. Not considering consumption of homegrown produce because there are
no current gardens is contrary to the workplan. Further, comments regarding houses that
burned down, are tax-forfeited or opted out of the interim action are irrelevant for risk
assessment. However, these comments are relevant for decision-making because they are
indicative of the fact that this entire area is blighted and that the population is
disadvantaged in many ways.
4.5.2.2, p. 4-137. The following statement illustrates the degree of data manipulation and
opacity of hazard and risk calculations in the HHRA: “For both tribal and standard
residents, the current cancer risk at the location with the minimum pre-IRM soil
concentrations (RES-05) is 8 percent of the pre-IRM value (Table 4-16). The current
cancer risk at the location with the maximum pre-IRM soil concentrations (RES-09) is 1
percent of the pre-IRM value.” Are we therefore to assume that cancer risks (using the
St. Regis HHRA comments
MDH, December 14, 2007
Page 20
old potency slope of 1.5 E+5/(mg/kg-d) are underestimated by factors between 12.5 and
100? These numbers highlight the lack of transparency in the risk calculations: based on
the dilution factors described in the HHRA one would not expect these pre-post
differences.
MDH suggests that EPA simply ignore IP’s calculations of cancer risks from exposure
point concentrations. EPA should go back to the original, pre-interim action data, and
calculate soil concentrations that yield cancer risks of 1E-6, 1E-5, 1E-4. This should be
done using the higher potency slope for reasons described above.
4.5.2.3, p. 4-138-140. Consumption of fish needs to be included in cancer risks from
Area B and combined Area A/B.
4.5.4, p. 4-142; 4.5.5, p. 4-145-146. The statement is made that non-dietary exposure
sources for the general population (soil, air and water), constitute only about 5 percent of
total dioxin exposure. The State of Minnesota default policy for evaluation of non-cancer
hazards from exposures to contaminated soil is that a relative source contribution of 20%
should be used. MPCA Soil Reference Values for non-cancer endpoints are thus based
on a hazard index of 0.2. Furthermore, IP notes that upper end background exposures for
the general adult population are about 1.1 pg/kg-d (above the ATSDR MRL); average
intakes are at 0.61 pg/kg-d (61% of the MRL). Childhood exposure is said to be as high
as 2.2 pg/kg-d. In the case of soil contaminated with dioxins, a good case can be made
that the remediation goal should be based on a hazard index below 0.2: perhaps 0.1 or
even 0.05.
4.5.5, p. 4-144. “The primary reservoirs, sediment and soil, act as environmental sinks
from which the highly persistent dioxins/furans compounds may be reintroduced into the
food chain.” This is a good argument for cleaning up the site.
4.6.1.1, p. 4-152. The following IP assumption is highly uncertain and not justifiable:
“EPCs for soil excluded subsurface soil data because the subsurface data set
is smaller than the surface data set and because subsurface sample locations
were selected in a biased fashion at locations with high surface
concentrations so that they probably overestimate typical subsurface
concentrations. At most locations where both surface and subsurface soil
samples were collected, surface concentrations are higher than subsurface
concentrations, so activities that mix surface and subsurface soil would tend
to decrease concentrations remaining at the surface.”
Conclusions and Recommendations
1. The interim actions should not be used in the risk assessment, per letters from
MDH (Patricia Bloomgren, 2003; Rita Messing and Daniel Peña, 2005).
2. Fish consumption should be included in the risk assessment.
St. Regis HHRA comments
MDH, December 14, 2007
Page 21
3. IP’s calculations cannot be used without extensive checking and modification.
4. EPA should complete a set of tables for cancer and non-cancer risks at different
concentrations of soil dioxins using agreed-upon exposure factors. These should
be used in conjunction with risks from exposures to sediments, groundwater,
surface water and fish.
5. To the extent that particular exposures are not quantitatively considered (e.g.,
surface deposition onto leafy plants, groundwater and surface water exposures,
eating fish eggs), these should be considered as uncertainties that will increase
calculated risks.
6. To the extent that gaps in sampling data cannot be quantitatively considered (e.g.,
lack of characterization of sub-surface soil, lack of data for PAH’s), these should
be considered as uncertainties that will increase calculated risks.
7. To the extent that general environmental factors are not considered (e.g., lack of
paved roads, and dusty environmental conditions), these should be considered as a
source of underestimation of exposures to soil and dust.
8. EPA should acknowledge that some parameters (e.g., bioavailability of dioxins in
dust particles, soil adherence factors, etc.) may be underestimated, and should be
considered as resulting in an underestimation of risk.
9. EPA should acknowledge that many individuals in the area are long term
residents, who may have had much larger exposures in the past. These exposures
probably occurred across generations.
10. EPA should acknowledge State of Minnesota and Tribal risk assessment policies
for assessment of cancer and non-cancer hazards and risks, and set remediation
goals to determine further site investigation and remedial actions that are
consisten with these policies.
11. EPA should act quickly to complete the risk assessment, establish conservative
remediation goals, and proceed to complete site investigation and cleanup.
MDH has repeatedly made these or similar comments. From the beginning, MDH
has sought to emphasize that this site is a blighted area on reservation land in the
middle of a small city with many tribal members and people living in poverty. It is an
urgent environmental justice issue and the timeline for action is entirely too long.
MDH also reminds EPA that State of Minnesota dioxin policies have guided the risk
assessment and cleanup of the Joslyn NPL site. EPA has used conservative criteria
for cleanup of the Escambia site in Florida and for assessing dioxin risks at the World
Trade Center. Use of State or Tribal policies and criteria for investigation and
cleanup of this site has ample precedent and is consistent with existing laws and
policies that guide EPA actions. In June 2007, John Linc Stine, Director of the
Division of Environmental Health wrote to Richard Karl, Director of the Superfund
Division, EPA Region 5 as follows:
“The situation for residents living on or near the site is little changed from what it was
in 1995.” Mr. Stine further went on to say: “It is time to implement a comprehensive
site investigation and remedial action at the site. Criteria for dioxin soil remediation
are available that are protective of public health. Minnesota has used a criterion of 35
St. Regis HHRA comments
MDH, December 14, 2007
Page 22
ppt for remediation of soils in industrial areas at the Joslyn site...(the Minnesota
criterion for residential soils is 20 ppt). EPA agreed to remediate soils at the
Escambia site in Pensacola, Florida using the Florida criteria of 30 ppt for industrial
areas and 7 ppt for residential areas.”
In March 2005, Dianne Mandernach, Commissioner of the Minnesota Department of
Health, wrote to Mr. Karl:
“Most of the people living in the impacted area are tribal members, many living in
poverty. As a result, we believe that this site poses an urgent environmental justice
concern.”
At the December 2006 National Environmental Public Health Conference in Atlanta,
GA, Dr. Howard Frumkin, Director, NCEH/ATSDR made the following remarks at
the opening plenary session
(http://www.cdc.gov/nceh/conference/NEHC%20Transcripts%20Edited_v2.pdf):
“Next, we need to pursue justice. We know across the public health world, including
in environmental public health, that not all of us are equally affected by public health
problems. Some people are disproportionately exposed and disproportionately at risk.
This has given rise to the transformative field of environmental justice. Poor
communities and communities of color have taught us all that people at risk deserve
special consideration.”
MDH letters to EPA, MDH’s dioxin risk assessment policies and the slide
presentations of Drs. Eaton and DeVito are attached for your convenience.
St. Regis HHRA comments
MDH, December 14, 2007
Page 23
Sincerely,
Patricia McCann
Rita Messing
Daniel Peña
Virginia Yingling
Minnesota Department of Health, Division of Environmental Health
References
1. Minnesota Pollution Control Agency (MPCA), Five-Year Review Report St. Regis Paper Company
Site Cass Lake Minnesota March 27, 1995.
2. U.S. EPA Region 5, Second Five-Year Review Report St. Regis Paper Company Site Cass Lake,
Cass County, Minnesota September 2000.
3. Third Five-Year Review Report St. Regis Paper Company Site Cass Lake, Cass County, Minnesota
September 2005.
4. U.S. EPA Region 5. Data Evaluation Report, for the St. Regis Paper Company Site, Cass Lake, Minnesota. August 23, 2002 5. U.S. EPA. Unilateral Administrative Order, V-W-'04-C-771 (Residential Soil Removal Action),
Pursuant to Section 10 CERCLA, Sept 19, 2005.
6. MPCA, Memorandum: Soil Reference Value Updates, September 7,2005
7. MDH Risk Assessment webpage, Methods for Estimating the Carcinogenic Health Risks from
Dioxin-Like Compounds (http://www.health.state.mn.us/divs/eh/risk/guidance/dioxinmemo1.htm l).
Updated October 2006
8. ATSDR, St. Regis Paper Company Health Consultation, August 2004.
9. MDH Letter to William Muno EPA Director Superfund Division: Removal Action Plan October 28,
2003.
10. MDH letter to Tim Drexler EPA Remedial Project Manager: Interim House Dust Plan. June 23, 2005. Attachment 1 MDH Dioxin Toxicity Assessment Policy
Risk Assessment: Dioxin Guidance: Minnesota Dept. of Health
Page 1 of 7
Risk Assessment
Rules/Guidance
The following guidance was developed by the Minnesota Department of Health (MDH) at the
request of the Minnesota Pollution Control Agency (MPCA). For more information, contact the
Health Risk Assessment Unit, 651/201-4899. For more information about dioxin health risks,
see Development of Inhalation Benchmark for Dioxin-Like Compounds and Facts about Dioxin.
Methods for Estimating the Carcinogenic Health Risks from
Dioxin­Like Compounds
Updated October 2006
The Minnesota Department of Health (MDH) prepared this guidance in response to a request in
2003 from the Minnesota Pollution Control Agency (MPCA), and to identify a consistent approach
for agencies and programs to assess the carcinogenic health risks from exposure to dioxin­like
compounds. Guidance for assessing the noncancer health risks is still under development and
will not be addressed in this memo. Because of the uncertainties associated with the toxicities of
dioxin mixtures, MDH uses a conservative approach to evaluate potential risks. As more data
become available, MDH re­evaluates and revises its risk assessment methods and procedures,
as appropriate.
Dioxin­like Compounds
The term "dioxins" is used to refer to a family of complex but related chlorinated compounds
with similar chemical structures and biological activity. The polychlorinated dibenzo­p­dioxins
(PCDDs) include 75 individual compounds, the polychlorinated dibenzofurans (PCDFs) include
135 individual compounds, and the polychlorinated biphenyls (PCBs) include 209 individual
compounds. These individual compounds are technically referred to as congeners. Based on
their ability to bind to the Ah receptor and evoke a response 7 of the 75 PCDD congeners (i.e.,
those with chlorine substitutions in the 2,3,7, and 8 positions), 10 of the 135 PCDF congeners
(i.e., those with chlorine substitution in the 2,3,7, and 8 positions), and 12 of the 209 PCB
congeners (those containing 4 or more chlorines with 1 or no substitutions at the ortho position)
are thought to have significant dioxin­like toxicity. The 29 compounds identified as having
significant dioxin­like toxicity concerns are identified in Table 1.
Toxic Equivalence Factors (TEFs)
Dioxins interfere with a basic and ubiquitous receptor system (the Ah receptor) that regulates
enzymes and other proteins. While it is believed that these 29 compounds have a similar
mechanism of toxicity not all are equally toxic. The most toxic and best­studied dioxin is
2,3,7,8­tetrachlorodibenzo­p­dioxin (2,3,7,8­TCDD). The remaining 28 compounds have been
assigned toxicity values relative to 2,3,7,8­TCDD. These relative toxicity values are called
toxicity equivalence factors (TEFs). 2,3,7,8­TCDD is assigned a TEF of 1 and the remaining
compounds are typically assigned values less than 1. The Minnesota Department of Health
(MDH) recommends utilization of the World Health Organization's (WHO) 2005 TEF scheme
(TEFWHO05) (Van den Berg, et al., 2005) to weight each compound according to its relative
toxicity for cancer risk evaluations. The TEFWHO05 values are shown in Table 2.
http://www.health.state.mn.us/divs/eh/risk/guidance/dioxinmemo1.html
11/30/2007
Risk Assessment: Dioxin Guidance: Minnesota Dept. of Health
Page 2 of 7
The compound specific TEF describes an order of magnitude consensus estimate derived from
scientific judgment based on the examination of available experimental data. TEF estimates are
generated from several sources of experimental data. The resulting range of relative potency
values derived from the individual experiments for a particular compound are variable. The TEFs
were primarily derived from in vivo toxicity data, which were given more weight than the in vitro
and/or quantitative structure­activity relationship data. The 2005 World Health Organization Re­
evaluation of Human and Mammalian Toxic Equivalency Factors for Dioxins and Dioxin­like
Compounds (Van den Berg et al. 2005) should be referred to for additional information on the
determination and validation of the TEFsWHO05.
The United States Environmental Protection Agency (U.S. EPA) as well as several states,
countries and international agencies have adopted the WHO 2005 TEF scheme. Utilization of the
WHO 2005 TEF scheme will facilitate the comparison of environmental measurements to
national and international databases.
Toxic Equivalent Concentration Calculation (TEQ)
The dioxin­like compounds exist in the environment as mixtures (i.e., a single compound is not
found in isolation). Because dioxins differ in their toxic potential or potency, the toxicity of each
component of the mixture must be accounted for in estimating the overall toxicity of the
mixture. The evaluation of environmental dioxin mixtures consists of three simple steps. The
first is a laboratory measurement of the concentration of each individual compound. Then, the
measured concentration of each compound is multiplied by its corresponding TEF to produce a
TCDD toxicity equivalent (TEQ) concentration. Finally, the TEQ concentrations for each
compound are added together with the TEQs for each of the other compounds present to
determine the total TCDD TEQ concentration in the sample. The total TCDD TEQ concentration
represents the amount of 2,3,7,8­TCDD alone, that it would take to equal the combined toxic
effect of the mixture.
The variability in relative potency values for individual compounds mentioned above may not
significantly impact an individual risk estimate. According to the U.S. EPA draft reassessment
only 5 compounds (2,3,7,8­TCDD, 1,2,3,7,8­PentaCDD, 1,2,3,6,7,8­HexaCDD, 2,3,4,7,8­
PentaCDF, and PCB 126) account for 70 to 80 percent of the TCDD TEQ in the human body and
food products. The relative potency variability reported in the literature for these 5 compounds
is much lower than for other compounds (U.S. EPA 2000).
A number of studies have examined the toxicity of complex mixtures of dioxins and non­dioxin­
like compounds in the laboratory. Some of these studies have compared the predicted (TEQ)
toxicity of a mixture to the actual measured toxicity of that mixture. Other studies have
compared the toxicity of individual compounds to those of the mixture in the same test system.
Mixtures tested include both laboratory mixtures of individual compounds and environmental
samples.
The TEF/TEQ methodology addresses the toxicity potential of complex mixture in terms of an
equivalent mass of 2,3,7,8­TCDD. Although the various congeners of a mixture have relative
equivalent toxicity to 2,3,7,8­TCDD, these congeners may not have the same pharmacokinetics
and do not necessarily share the same environmental fate as 2,3,7,8­TCDD. The impact of an
environmental mixture will likely also be affected by the ability of our bodies to absorb,
metabolize and excrete the individual congeners from the environmental media (e.g. soil). For
some risk assessments the differences in fate and transport of different congeners must be
taken into consideration and TEQs calculated at the point of exposure to achieve more accurate
assessments.
http://www.health.state.mn.us/divs/eh/risk/guidance/dioxinmemo1.html
11/30/2007
Risk Assessment: Dioxin Guidance: Minnesota Dept. of Health
Page 3 of 7
Although the use of TEFs and the TEQ approach is widespread, its use is not without
controversy. The WHO has suggested that the TEF scheme and the TEQ methodology be re­
evaluated every 5 years to account for new scientific information. The WHO completed their
most recent review of the TEFs and TEQ methodology in 2005 (Van den Berg et al., 2005). In
this review they reaffirmed the use of TEFs as the best available tool for estimating the health
risk of exposures to complex mixtures of dioxin­like chemicals.
Quantitative Cancer Risk Assessment
Cancer Classification:
The MDH, U.S. EPA, National Toxicology Program (NTP) and the International Agency for Cancer
Research (IARC) have characterized 2,3,7,8­TCDD as a "human carcinogen". The MDH and the
U.S. EPA have classified the complex mixtures of dioxin to which people are exposed as a "likely
human carcinogen". The degree of certainty of the cancer hazard is dependent on the major
constituents of the mixture. The consistent, suggestive evidence from epidemiology studies
combined with the unequivocal evidence in animal studies and inferences drawn from
mechanistic data support the characterization of complex mixtures of dioxin and related
compounds as "likely" cancer hazards. "Human carcinogen" and "likely" are descriptors which
are consistent with the U.S. EPA draft final cancer guidelines (U.S. EPA 2003). They are roughly
equivalent to the terms "known" and "probable" human carcinogen contained in earlier U.S. EPA
cancer guidelines (U.S. EPA 1986).
Oral Cancer Slope Factor:
The U.S. EPA's draft dioxin reassessment efforts produced two upper bound slope factors for
estimating human cancer risk from exposure to dioxins:
1 x 10­3 (pg TCDD TEQ/ kg body weight/day)­1 based on an evaluation of the human
epidemiology data and 1.4 x 10­3 (pg TCDD TEQ/kg body weight/day) ­1 based on a re­
evaluation of the animal data (liver cancer in female rats).
The actual shape of the low­dose exposure­response relation for animals or humans cannot be
determined from the available data. For this reason U.S. EPA utilized a linear dose extrapolation
model to derive an upper bound cancer potency factor. The true risk is unknown but is likely to
be lower.
The MDH believes that exposure to ("known" or "likely") carcinogens should be minimized where
possible; this is especially true for dioxins due to existing body burden estimates. When a
numerical cancer slope factor is needed to evaluate incremental risk, MDH recommends utilizing
an interim cancer slope factor of 1.4 x 10­3 (pg TCDD TEQ/kg body weight/day) ­1 (i.e., 1.4 x
106 per mg TCDD TEQ/kg/day). This value is based on EPA's draft animal­based cancer slope
factor. Concerns about the quality of the exposure estimates in the human epidemiological
studies preclude the quantitative use of these data in developing a cancer potency slope for
dioxin; however, the results from modeling the human studies are consistent with the cancer
potency slope derived by modeling data from animal studies.
As noted above, the recommended interim cancer slope factor is based on information contained
in the EPA 2003 draft dioxin reassessment. The National Research Council (NRC) of the National
Academies has recently completed a comprehensive review of the EPA 2003 draft assessment.
The NRC report contained conclusions and recommendations on how the EPA 2003 draft
reassessment could be improved. The MDH is initiating a review of the NRC report as well as
other relevant scientific data generated since the draft reassessment (e.g., the chronic
http://www.health.state.mn.us/divs/eh/risk/guidance/dioxinmemo1.html
11/30/2007
Risk Assessment: Dioxin Guidance: Minnesota Dept. of Health
Page 4 of 7
toxicological study conducted by the National Toxicology Program). The MDH does not
recommend any changes to its guidance at this time.
The recommended slope factor is derived from the same study, Kociba et al., 1978, as the
previous slope factor estimate (1.56E+5 per mg/kg/day). The development of the
recommended slope factor utilized current methods of analysis, including the use of body
burden as the dose metric for animal­to­human dose equivalence calculations (i.e., adjustments
to account for the differences in half­life of dioxins in the bodies of laboratory animals and
humans), and a re­evaluation of the liver tumors in the Kociba study using the latest pathology
criteria.
Inhalation Unit Risk Factor:
For bioaccumulative compounds such as the dioxins the primary exposure route of concern for
long­term or chronic toxicity is ingestion rather than inhalation. Toxicological data from
inhalation studies is not available for the dioxins. As stated in the MDH Statement of Need and
Reasonableness (SONAR) for the Health Risk Values (HRVs) chronic ingestion studies are not, as
a rule, utilized by the MDH to develop inhalation HRVs. Route­to­route extrapolation may be
appropriate when sufficient toxicokinetic information is available and the critical effect would be
the same regardless of how the toxicant is administered.
There is adequate evidence, in laboratory animals, that 2,3,7,8­TCDD is a multisite carcinogen
capable of increasing the incidence of tumors at sites distant from the site of treatment. The
limited epidemiologic evidence from occupationally exposed workers is also consistent with
increased cancer risk at multiple sites.
The situations where extrapolation from an oral exposure to an inhalation exposure is
inappropriate are also discussed in the SONAR:
"There are, however, situations where this extrapolation technique is inappropriate. For
instance, if the critical effect is specific for the respiratory system, or if the toxicity of a chemical
is expressed at or near the site of application, data from oral exposure should not be used to
extrapolate to an inhalation exposure. Another case where extrapolation would be inappropriate
is when the target organ for the critical effect is the liver. The liver, because of its unique
structure and circulation, is subjected to much higher concentrations of ingested chemicals than
other organs. In addition, the unique biochemistry of the hepatocytes can result in the
generation of very different metabolic products of a toxicant in the liver than would be produced
in other organs. For these reasons an extrapolation approach will not be used if the liver is the
target organ for a toxicant following oral exposure."
The liver is a target organ of dioxin toxicity and the recommended oral slope factor is based on
liver tumors. The MDH HRV staff were consulted regarding interpretation of the SONAR and the
appropriateness of utilizing route­to­route extrapolation for dioxins. 2,3,7,8­TCDD and dioxin­
like compounds undergo limited metabolism and exhibit long half­lives in the body. As a result
the liver would not be subjected to significantly higher concentrations or significantly different
metabolic products than other organs. Therefore, although the recommended oral slope factor is
based on liver tumors, route­to­route extrapolation is acceptable.
In order to extrapolate an inhalation unit risk factor from the oral slope factor an absorption
adjustment factor, inhalation rate, and body weight are necessary. These parameter values are
influenced by the physical form of dioxins (e.g., particulate or vapor­phase) and the individual
or population under evaluation. MDH will not recommend default adjustment factors, inhalation
rates or body weights at this time. As more data become available, MDH will re­evaluate this
http://www.health.state.mn.us/divs/eh/risk/guidance/dioxinmemo1.html
11/30/2007
Risk Assessment: Dioxin Guidance: Minnesota Dept. of Health
Page 5 of 7
position and revise its recommendation as appropriate.
Information contained in the U.S. EPA draft dioxin reassessment may be a useful source of
information for the MPCA. Part I, Volume 3, Chapter 4 and Volume 4, Chapter 2 of the U.S.
EPA's draft reassessment provide guidance for estimating potential risks from a variety of
exposure pathways, including the inhalation pathway.
Utilizing animal data and information on fate of particles in the respiratory system, U.S. EPA
estimated that the fraction of 2,3,7,8­TCDD absorbed into the body ranges from 0.25 to 0.29.
Although the rate of absorption of vapor­phase 2,3,7,8­TCDD into the lungs has not been
studied, the U.S. EPA concluded that it seems reasonable to assume that the absorption in the
vapor phase should exceed that of absorption from bound 2,3,7,8­TCDD on particulates,
probably above 50%. Given the paucity of data U.S. EPA recommended that assessors not
attempt any such adjustments at this point, but fully acknowledged the uncertainty. Absorption
correction factors were recommended for use in the soil ingestion and soil dermal contact
pathways.
A variety of inhalation rates and body weights were utilized by the U.S. EPA to estimate
inhalation exposure. The specific value selected depended on the age of the subpopulation of
concern and whether a central tendency or upper tendency estimate was desired.
Table 1. List of Compounds With Varying Dioxin­like Toxicity
Chlorinated Dibenzo­p­
dioxins (CDDs)
2,3,7,8­TCDD
1,2,3,7,8­PentaCDD
1,2,3,4,7,8­HexaCDD
1,2,3,6,7,8­HexaCDD
1,2,3,7,8,9­HexaCDD
1,2,3,4,6,7,8­HeptaCDD
1,2,3,4,6,7,8,9­OctaCDD
Chlorinated
Dibenzofurans (CDFs)
2,3,7,8­TetraCDF
1,2,3,7,8­PentaCDF
2,3,4,7,8­PentaCDF
1,2,3,4,7,8­HexaCDF
1,2,3,6,7,8­HexaCDF
2,3,4,6,7,8­HexaCDF
1,2,3,7,8,9­HexaCDF
1,2,3,4,6,7,8­HeptaCDF
1,2,3,4,7,8,9­HeptapCDF
1,2,3,4,6,7,8,9­OctaCDF
Polychlorinated Biphenyls
(PCBs)
3,3'4,4'­TetraCB (PCB 77)
3,4,4',5­TetraCB (PCB 81)
2,3,3',4,4'­PentaCB (PCB 105)
2,3,4,4',5­PentaCB (PCB 114)
2,3',4,4',5­PentaCB (PCB 118)
2',3,4,4',5­PentaCB (PCB 123)
3,3',4,4',5­PentaCB (PCB 126)
2,3,3',4,4',5­HexaCB (PCB 156)
2,3,3',4,4',5'­HexaCB (PCB 157)
2,3',4,4',5,5'­HexaCB (PCB 167)
3,3',4,4',5,5'­HexaCB (PCB 169)
2,3,3',4,4',5,5'­HeptaCB (PCB 189)
Table 2. WHO05 Toxic Equivalent Factors
Compound
TEFWHO05
CDDs
2,3,7,8­TetraCDD
1,2,3,7,8­PentaCDD
1,2,3,4,7,8­HexaCDD
1,2,3,6,7,8­HexaCDD
1,2,3,7,8,9­HexaCDD
1,2,3,4,6,7,8­HeptaCDD
1,2,3,4,6,7,8,9­OctaCDD
CDFs
1
1
0.1
0.1
0.1
0.01
0.0003
http://www.health.state.mn.us/divs/eh/risk/guidance/dioxinmemo1.html
11/30/2007
Risk Assessment: Dioxin Guidance: Minnesota Dept. of Health
Page 6 of 7
2,3,7,8­TetraCDF
0.1
1,2,3,7,8­PentaCDF
0.03
2,3,4,7,8­PentaCDF
0.3
1,2,3,4,7,8­HexaCDF
0.1
1,2,3,6,7,8­HexaCDF
0.1
2,3,4,6,7,8­HexaCDF
0.1
1,2,3,7,8,9­HexaCDF
0.1
1,2,3,4,6,7,8­HeptaCDF
0.01
1,2,3,4,7,8,9­HeptaCDF
0.01
1,2,3,4,6,7,8,9­OctaCDF
0.0003
PCBs
3,3'4,4'­TetraCB (PCB 77)
0.0001
3,4,4',5­TetraCB (PCB 81)
0.00003
2,3,3',4,4'­PentaCB (PCB 105)
0.00003
2,3,4,4',5­PentaCB (PCB 114)
0.00003
2,3',4,4',5­PentaCB (PCB 118)
0.00003
2',3,4,4',5­PentaCB (PCB 123)
0.00003
3,3',4,4',5­PentaCB (PCB 126)
0.1
2,3,3',4,4',5­HexaCB (PCB 156)
0.00003
2,3,3',4,4',5'­HexaCB (PCB 157)
0.00003
2,3',4,4',5,5'­HexaCB (PCB 167)
0.00003
3,3',4,4',5,5'­HexaCB (PCB 169)
0.03
2,3,3',4,4',5,5'­HeptaCB (PCB 189) 0.00003
References
CalEPA. California Environmental Protection Agency, Office of Environmental Health Hazard
Assessment (OEHHA). Air Toxics Hot Spots Program Risk Assessment Guidelines, Part II:
Technical Support Document for Describing Available Cancer Potency Factors, December 2002.
Chlorinated Dibenzo­p­Dioxins and Chlorinated Dibenzofurans, pp. 167 ­ 187.
http://www.oehha.ca.gov/air/hot_spots/pdf/TSDNov2002.pdf
Minnesota Department of Health (MDH) Statement of Need and Reasonableness, Proposed
Permanent Rules Relating to Health Risk Values, Minnesota Rules, Parts 4717.8000 to
4717.8600, August 10, 2001.
NTP. Report on Carcinogens, Tenth Edition: U.S. Department of Health and Human Services,
Public Health Service, National Toxicology Program, December 2002.
http://ehp.niehs.nih.gov/roc/toc10.html
U.S. EPA, 1986. United States Environmental Protection Agency. Guidelines for Carcinogen Risk
Assessment. Published in the Federal Register 51(185): 33992 ­ 34003, September 24, 1986.
http://www.epa.gov/ncea/raf/car2sab/guidelines_1986.pdf
U.S. EPA, 2000. United States Environmental Protection Agency. Draft Exposure and Human
Health Reassessment of 2,3,7,8­Tetrachlorodibenzo­p­dioxin TCDD) and Related Compounds.
Draft Final, September 2000. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=55265
U.S. EPA, 2003. United States Environmental Protection Agency. Draft Final Guidelines for
Carcinogen Risk Assessment. February 2003. EPA/630/P­03/001A.
http://www.health.state.mn.us/divs/eh/risk/guidance/dioxinmemo1.html
11/30/2007
Risk Assessment: Dioxin Guidance: Minnesota Dept. of Health
Page 7 of 7
http://www.epa.gov/ncea/raf/cancer2003.htm
Van den Berg, M, L. Birnbaum, et al., The 2005 World Health Organization Re­evaluation of
Human and Mammalian Toxic Equivalency Factors for Dioxins and Dioxin­like Compounds.
Toxicological Sciences, October 2006; 93:223­241
WHO, World Health Organization. Executive Summary ­ Assessment of the Health Risks of
Dioxins: Re­evaluation of the Tolerable Daily Intake (TDI), WHO Consultation, May 25 ­ 29,
1998, Geneva, Switzerland. http://www.who.int/fsf/Chemicalcontaminants/whoinfo.htm
For questions about this page, please contact our Environmental Health Division: [email protected]
Updated Wednesday, 14­Mar­2007 11:57:55 CDT
http://www.health.state.mn.us/divs/eh/risk/guidance/dioxinmemo1.html
11/30/2007
Attachment 2 MDH Letters to EPA Superfund Division Director
Attachment 3 Dr. Eaton’s NAS/NRC Report: Health Risks From Dioxins and Related Compounds NAS/NRC Report:
Health Risks from Dioxin and Related
Compounds
The NAS and WHO on Dioxin and
Dioxin Like Compounds:
International Policy Implications and
Potential Impacts
Michigan State University
Superfund Program Workshop
David L. Eaton, Ph.D.
Professor, Dept. Env. Occupational
Health Sciences
Associate Vice Provost for Research
University of Washington
Dioxins, Dibenzofurans and PCBs
•
•
•
Chlorinated Dioxins represent a class of compounds, of which 7 are
included in EPA regulations (Contaminants – no commercial use)
Chlorinated dibenzofurans include 10 congeners (formed from PCBs)
Certain ‘planer’ PCBs also have ‘dioxin-like’ activity, and are included.
Toxicology of ‘Dioxin’ in 1 Minute
• Very toxic, both acute and chronic
– LD50 0.6 ug/kg in sensitive species
– Chronic – birth defects, immunotoxicity, cancer, chloracne,
reproductive effects, liver, CNS
– Large species differences
• Mechanism of all (or nearly all) toxic effects is by
binding to the Ah Receptor
– Transcriptional activation of numerous genes, especially
CYP1A1
– Toxic effects are result of ‘downstream’ events that follow Ah
Receptor activation
• Very fat soluble, resistant to degradation
– Persistent in the environment
– Bioaccumulate
– Long biological half life (about 6-7 yrs in humans)
Known Effects of TCDD in Humans
Before Dioxin poisoning
After Dioxin poisoning
Yuschenko’s blood TCDD concentration was ~ 100,000 pg / g lipid
Known Effects of DLCs in Humans
• Yusho and Yu-Cheng Disease in Japan
– Massive PCB/PCDF exposures via diet
• Michigan nursing mothers exposed to PCBs via fish
consumption
• Dutch cohort of off-spring of moderately exposed mothers
(DLCs, mostly PCBs)
– Developmental abnormalities following in utero exposures, mostly
neurobehavioral and cognitive deficits
– Some evidence of immune system dysfunction
– Some evidence of hormonal/endocrine dysfunction
• Occupational exposures – possible increase in cancers
Dioxins, Diobenzofurans & PCBs
Toxic Equivalence Factors (TEFs)
History – EPA Dioxin Risk Assessment
•
1985 – Completed first risk assessment of Dioxin
• Classified as potent, ‘likely’ human carcinogen
• linear extrapolation, mg/kg-d dose metric
• Controversial assumptions
•
•
1991 – EPA announces that it will Reassess Dioxin Risks
1994 – First draft of Reassessment released
• Basically supported findings of 1985 assessment
• Used body burden as dose metric
• Extensive peer review and public comments raised concerns about
models and assumptions
•
2000 –Revision of 1994 Reassessment released
• Additional Peer review and SAB comments
• Questions about linear model for cancer
•
2003 – Revised ‘Near Final’ Reassessment
• Requested the National Academies to do review of Reassessment
•
•
2004 – NAS/NRC appoints panel, begins review (Nov, 04)
2006 – NAS/NRC report released (July)
Why is this important?
Policy implications
• Many industries, (pulp and paper, chemical
manufacturers, incinerators, etc), have dioxin emissions
that are regulated
– Emissions standards will be based on risk
• Many state and federal (Superfund) hazardous waste
sites contain dioxins/DLCs
– clean-up standards will be based on risk
– State agencies can set own standards (if stricter than feds)
• Draft Reassessment suggested that there was potentially
unacceptable cancer risks at current background levels
– Implications for regulation of foods, especially meat and dairy
• Some regulations have been ‘on hold’ pending
acceptance of a final EPA Reassessment
The NAS/NRC Process
Selecting the Committee
• Committee of ‘highly respected’ scientists, with all
relevant areas of expertise represented
– Not necessarily ‘experts’ on dioxin, but have high level of
credibility in their discipline
• Full disclosure of potential conflicts and biases
• Not involved in preparation or review of the EPA
Reassessment
• Committee selected by the NAS leadership, following
detailed ‘vetting’ of information on nominees
• Tentative Committee becomes ‘final’ committee after 1st
meeting, when bias and conflict of interest are discussed
End Result – Committee of 18
•
•
•
•
•
•
•
•
•
•
Dave Eaton, PhD, UW (Chair)
Dennis Bier, MD, Baylor
Joshua Cohen, PhD, Harvard
•
•
(now at Tufts)
•
Mike Dennison, PhD, UC-Davis
Rich DiGiulio, PhD, Duke
Norb Kaminski, PhD, Mich. St.
Nancy Kim, PhD, NY St DoH
Djien Liem, PhD, European
Food Safety Authority, Italy
Tom McKone, PhD, UC-B
Malcolm Pike, MD, USC
Alvaro Puga, PhD, U Cinn.
Andy Renwick, PhD, Univ.
Southampton, UK
David Savitz, PhD, UNC
(now at Mt. Sinai)
•
•
•
•
•
Allen Silverstone, PhD, SUNYUpstate (Syracuse)
Paul Terranova, MD, KUMC
Kim Thompson, PhD, Harvard
(now at MIT)
Gary Williams, MD, NYMC
Yilang Zhu, PhD, U. S. Florida
Members highlighted in blue have spent much of their careers on dioxin toxicology
Committee Charge
The NAS/NRC will convene an expert committee that will review
EPA’s 2003 draft Reassessment….to assess whether:
• Risk estimates are scientifically robust
• There is a clear delineation of all substantial uncertainties
and variability
• To the extent possible, focus on:
– Modeling assumptions (shape of D-R curve, points of departure,
dose ranges for likely human health outcomes)
– EPA’s quantitative uncertainty analysis
– EPA’s selection of studies as a basis for its assessments
• Also address:
– Scientific evidence classifying TCDD as a human carcinogen
– Validity of the non-threshold, linear D-R model and slope factors
– Usefulness of TEF/TEQ approach
Other Conditions / Limitations
• Complete the review in 18 months
– including peer review, revisions, and final editing
•
•
•
•
•
Solicit public input prior to writing report
Strive for a ‘Consensus’ report
Have no more than 5-6 meetings
Draft report subject to extensive peer review
Final report must include consideration of all peer
review comments
• We focused our on review Part III of the Reassessment
– “Integrated Summary and Risk Characterization”
– ~200 page summary of the several thousand page report
• Considered ‘new information’ only if it was critical to key
assumptions, and likely to change the RA
Process
•
2 meetings with invited presentations
–
–
•
Heard from 16 different ‘interested parties’
Received piles of solicited and unsolicited information
Organized the report to include 8 chapters:
1) Introduction
2) General Considerations of Uncertainty and Variability, Selection of Dose
Metric, and Dose-Response Modeling
3) Toxic Equivalency Factors
4) Exposure Assessment
5) Cancer
6) Non-Cancer Endpoints (Immune, Repro/Development, Other)
7) Risk Characterization
8) Conclusions and Recommendations
•
•
•
•
Chapter review assignments made by Chair, based on areas of
expertise – 3-4 members per topic
2 meetings to discuss recommendations
1 meeting to finalize report conclusions and consensus
Most of the real work was done by e-mail
Process (cont)
•
•
•
•
Consensus draft report completed in December, 2005
Edited by staff then sent to 15 different peer reviewers
3 months later, received ~120 pages of comments
Made numerous changes (requiring approval of all
committee members) and submitted revised to
NAS/NRC Study Monitor
– along with detailed list of how document was changed in
response to comment, and if not, why not
• After approval of Study Monitory, Final draft report sent
to NRC staff for editing and printing
• Congressional Briefings and Press Conference held on
day before report was released (July 11, 2006).
Key Findings of The Committee
• “3 areas that require substantial improvement in describing
the scientific basis for EPA’s dioxin risk assessment”
– Justification of approaches to D-R modeling
– Transparency and clarity in selection of key data sets
– Transparency, thoroughness and clarity in quantitative uncertainty
analysis
• Classification of TCDD as known vs. likely human carcinogen
– Use the new definitions in 2005 CAG
• TEFs continue to be best approach for assessing mixtures
• Encouraged EPA to calculate RfD and MOE scenarios
Key issue – Qualitative Assessment of
TCDD carcinogenicity to humans
• Seems to be a ‘big deal’ to lots of people
• Committee felt that it really was not important, as TCDD will
(and should) be regulated as if it is carcinogenic to humans
regardless of what label it is given
• Better off spending time on more critical uncertainties that will
affect the quantitative risk estimations
• Guidelines and definitions changed in 2005 – EPA should use
new guidelines, then justify their decision
• Committee was ‘split’ on whether the available data met the
criteria of the new guidelines
– Full agreement that it was at least likely to be carcinogenic in humans
– Other DLC congeners ‘Likely’ to be carcinogenic to humans
“Carcinogenic to Humans”
EPA Carcinogen Assessment Guidelines 2005
This descriptor indicates strong evidence of human carcinogenicity.
This descriptor is appropriate when there is convincing epidemiologic
evidence of a causal association between human exposure and cancer.
Exceptionally, this descriptor may be equally appropriate with a lesser
weight of epidemiologic evidence that is strengthened by other lines of
evidence. It can be used when all of the following conditions are met:
(a) there is strong evidence of an association between human exposure and
either cancer or the key precursor events of the agent's mode of action but
not enough for a causal association, and
(b) There is extensive evidence of carcinogenicity in animals, and
(c) the mode(s) of carcinogenic action and associated key precursor events
have been identified in animals, and
(d) there is strong evidence that the key precursor events that precede the
cancer response in animals are anticipated to occur in humans and progress
to tumors, based on available biological information.
Epidemiological Evidence
• Four occupational cohorts with substantial TCDD exposure
–
–
–
–
–
Ott & Zober, 1996 – 1953 accidental exposure (N=243, 13 cancer deaths)
Becher et al 1998 – Pesticide production cohort (N=1189; 124 Ca deaths)
Fingerhut et al (’90, 91) – 12 manufacturing facilities (N=5172; 377 deaths)
Steenland et al (2001) – update on Fingerhut cohort (N=3538; 256 deaths)
De Mesquita et al (1993) – Phenoxy production (N=2310, 31 cancer deaths)
• Most, but not all, found significant increase in all cancers,
but no consistent increase in any specific tumor type
• Committee conclusions:
– Overall, the committee concurs with the value of conducting
analyses of total cancers given the potential for dioxin to affect
multiple types of cancer
– “It was the Committee’s impression that EPA’s narrative tended to
focus on positive findings without fully considering the strengths and
limitations of both positive and negative findings.”
Key Issue – Shape of the D-R Curve
•
•
•
•
•
Mode of action – Receptormediated for all end points
Non-genotoxic
Evidence of tumor promotion
Binding to Ah receptor is
necessary, but not sufficient, to
cause cancer
Existing animal and human epi
data provide little guidance as
to the shape of the D-R at
response levels below 5-10%
Rationale for EPA’s Choice of Linear,
non-threshold model
•
•
•
•
•
“At this time, the knowledge of the mechanism of action of dioxin,
receptor theory, and the available dose-response data do not firmly
establish a scientific basis for replacing a linear procedure for
estimating cancer potency.”
“The linear default is selected on the basis of the agent’s mode of
action when the linear model cannot be rejected and there is
insufficient evidence to support an assumption of non-linearity”
Committee disagreed with using the ‘default’ assumption, given the
enormous amount of data on dioxin mode of action, and noted that
EPA had used non-linear modeling for other receptor-mediated
carcinogens (thyroid carcinogens, estrogens, etc.)
Recommended that they do BOTH, to illustrate the importance of this
assumption
If they choose to use the linear estimates out of ‘precaution’, that
would be a policy decision, with the implications clearly described
Key Issue – Point of Departure
• EPA used a POD of 1%: “The curve-fitting procedure is
used to determine a POD, generally at the 10% response level,
but when more sensitive data are available, a lower point for
linear extrapolation can be used to improve the assessment
(e.g., 1% response for dioxin, ED01).
• They calculated 1% PODs from the epidemiology
data, using various models
• Use of ED05 would greatly expand the Confidence
Limits around the central estimate (from which the
Slope factors are derived)
• “It is evident that the choice of POD can have a
substantial impact on the uncertainty of the final risk
estimate – importance of this assumption is not
readily evident in the Reassessment””
EPA Modeling of Cancer Data for ED01
Hamburg Cohort
NIOSH cohort
Mean Serum TCDD: 507 (2 – 6397 ppt)
Mean Serum TCDD: 2000 (2 – 32000 ppt)
EPA Modeling of Cancer Data for ED01
BASF cohort
Mean Serum TCDD: 1008 (20 – 13360 ppt)
Male Sprague-Dawley Rat data
Cancer Slope Factors Derived from
ED01 modeling
EPA Conclusions for Cancer Slope Factors
Main concern of Committee
• Significance of ED01 vs. ED05 for POD
• Alternative, biologically plausible, dose-response
functions
• Final cancer slope factor estimates from Epi studies
ranged from 0.9 x 10-3 to 5.1 x 10-3 (6-fold) and
compared with two estimates from rats data of 0.8 x
10-3 and 0.97 x 10-3. (all within a factor of 10).
• Committee felt that the range of uncertainty is greater
than indicated in Reassessment
NRC Report
Range of Plausible CSF Values –
Consideration of Parameter Confidence Intervals Only
CSF - Risk per pg/kg-day per
Million Population
100000
10000
1000
100
10
1
EPA
Ott and Zober
Becher
Data Source
Steenland
Key Issue – Dose Metric
• EPA used ‘body burden’ rather than daily intake rate
(pg/kg/day)
• Makes a very substantial difference (~280-fold) in
cancer risk estimates from animal studies because of
species differences in half-lives (~100-fold) and body
fat composition, and thus the daily dose that yields a
particular dioxin body burden at steady state
• Committee agreed with EPA that body burden,
although not perfect, is the best dose metric to use
for TCDD and DLCs, given their long half-lives and
bioaccumulation in adipose tissue.
Key Issue – Use of TEF/TEQ
– Use new TEFs from WHO
– Encourage development of stronger scientific basis
for individual congener TEFs, esp. those that are
‘drivers’
– Background levels of dioxins in environment are
declining- are body burdens also?
– Most of the body burden is a result of a few
congeners, and little or no TCDD
• EPA Reassessment used a ‘peak’ TEQ value of 55
pg/g lipid (30 – 70 CLs) as median US Background,
and 5.2 +/- 1.3 pg/g lipid for TCDD -- from 1990’s
LOD = 5.8 pg/g lipid
LOD = 9.1 pg/g lipid
Background serum levels of Dioxin TEQs in
NHANES II (US, 2001)
Assumes that non-detects are Zero
Sum of GM for all detects = 6.4 pg/g lipid
60.00%
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%
OCDD
HpCDD
HxCDD
HpCDF
PCB 126
PCB 169
1,
C
2,
on
3,
4,
ge
6,
ne
7,
1,
8
r
2,
-H
3,
pC
4,
D
7,
1,
D
8
-H
2,
3,
xC
6,
D
7,
D
1,
8
2,
-H
3,
xC
7,
D
8,
D
9
-H
1,
2,
xC
3,
D
7,
D
8Pe
2,
C
3,
D
D
7,
1,
82,
TC
3,
4,
D
7,
D
1,
8
2,
-H
3,
xC
6,
D
7,
F
1,
8
-H
2,
3,
xC
7,
D
8,
F
2,
9
-H
3,
4,
xC
6,
D
7,
F
8
2,
-H
3,
xC
4,
D
7,
F
8Pe
2,
C
3,
D
F
7,
8TC
D
PC F
B
12
PC 6
B
16
9
NHANES II Dioxin TEQ Data
Geometric Mean, using 0.5 x DL for non-detects
Sum of GM = 16.3 pg/gm lipid; Sum of 95% = 60.8 pg/gm lipid
25%
20%
15%
10%
5%
0%
Key Issue- Calculation of RfD
• EPA chose not to calculate a Reference Dose or
Margins of Exposure, primarily because their risk
estimates would have resulted in low (or negative)
MOEs.
• However, EPA’s approach for RfD/MOE calculations
use a number of uncertainty factors (such as a factor
of 10 going from rodents to humans), which may not
be necessary if Body Burden is used as dose metric
• WHO and many other countries utilize this approach
Bottom Line implications
• EPA’s draft Reassessment arrived at a cancer
slope factor of 1 x 10-3 per pg TEQ/kg/day
– If 1 excess cancer per 100,000 was used as the
‘acceptable risk’ level, this would result in an
‘Tolerable Daily Intake’ rate of 1 x 10-5 pg TEQ/kg/d,
or 0.01 pg TEQ/kg/d
– If 1 excess cancer per 1 million was used, the TDI=
0.001 pg/kg/d
• These risk levels are based on the linear
extrapolation assumption. Use of a non-linear
risk estimates would use a Benchmark dose and
Uncertainty factor approach
From Chlorine Chemistry Council
Next steps
• Although highly critical of a few key assumptions,
overall the Committee endorsed much of what was
done in the EPA Reassessment
• “Committee recognizes that iw will require a
substantial amount of effort for EPA to incorporate all
the changes recommended in the NRC report”
• “Nevertheless, the committee encourages EPA to
finalize the current Reassessment quickly, efficiently,
and concisely as possible after addressing the major
recommendations in the report”
Midland Daily News
1/04/07
Dioxin bill signed by Granholm
Legislation allowing the state to begin recalculating the dioxin
cleanup standards by incorporating the recommendations made by
the National Academy of Sciences was signed into law on Dec. 31,
2006.
"This legislation calls for the best available science to better protect
our health and our natural resources," Moolenaar said. "The
Legislature and governor have come together to support using
sound science for environmental cleanup, including the work
conducted by the independent National Academy of Scientists, to
lead to a more productive resolution."
Attachment 4 Dr. Devito’s presentation: Dose Response Relationships for Dioxins; Implications for Pubic Health Dose Response Relationships for Dioxins:
Implications for Public Health
Michael DeVito
Chief, Pharmacokinetics Branch
USEPA
Office of Research and Development
National Health & Environmental Effects Research Laboratory
October 15, 2007
Outline
• Dose
• Response Modeling
• Non-Cancer endpoints
• Cancer
• Summary
Office of Research and Development
National Health & Environmental Effects Research Laboratory
1
Dose Metrics
• A measure of dose that incorporates physical chemical
and/or biological attributes that relates exposure to a
biological response. A useful metric is easy to measure and
interpret
•
Exposure
– mg/kg; mg/kg/d; ppm
•
Tissue Concentration or Body Burden
– Peak
– Average lifetime
– Window of Sensitivity
•
Area Under the Curve (AUC)
– Lifetime
– Window of Sensitivity
– Time over a set concentration
Office of Research and Development
National Health & Environmental Effects Research Laboratory
2
Dose metrics used for Pharmaceuticals
• Administered Dose
• Plasma/Blood Concentrations
– Single dose: terminal elimination phase
– Repeated dose: Css
Blood/Plasma concentrations are related to
concentration of drug at target site
Office of Research and Development
National Health & Environmental Effects Research Laboratory
3
Do we need different dose metrics for environmental chemicals?
• Extrapolation
– Across species
– Across exposure scenarios/paradigms
– In vitro to in vivo
• Dose Response Assessment
– Cancer vs non-cancer
– Adult vs developmental toxicities
Office of Research and Development
National Health & Environmental Effects Research Laboratory
4
Methods for Cross-Species Dose Extrapolation in Risk Assessments
• Uncertainty Factors (10X)
– Animal to human
– Acute to chronic exposures
• Allometric scaling
– Biochemical and physiological processes scale across
species by functions of body mass or surface area
• “Mechanistic approaches”
– Physiologically-based pharmacokinetic models
– Systems Biology approach
Office of Research and Development
National Health & Environmental Effects Research Laboratory
5
Species Differences in Half-lives of Dioxin (TCDD)
Experimentally Estimated
Allometric Scaling
Half-life
(body weight) of
(days)
Mouse Half-life
(days)
Mice Rat
Rhesus
10
-
15-25
18.9
400
39.8
2593
76.9
Monkey
Humans
Office of Research and Development
National Health & Environmental Effects Research Laboratory
6
10000
1000
100
H
um
an
(1
.0
)
an
H
um
at
(
R
Office of Research and Development
National Health & Environmental Effects Research Laboratory
(0
.0
1)
10
1.
0)
Steady-state Body Burden
(ng/kg)
Steady-state Body burdens following
1 or 0.01 ng TCDD/kg/d exposure
7
Comparison of Lifetime Average Body Burden and Area under the Curve
in Hypothetical Background and Occupational Scenarios
LABB
AUC
n g /kg n g /kg * Y
P eak B B
B ack ground
3.6
O ccu p atio n al 5 5 .9 255
3911
L A B B = A U C /Y rs
L ifetim e A v e. B o d y
B u rd en O ccu p = 5 5 .9 n g /k g
L A B B = 3 .6 n g /k g
Office of Research and Development
National Health & Environmental Effects Research Laboratory
A rea U n d er th e C u rv e
- O ccu p atio n al
= 3 9 1 1 n g /k g * Y
A U C = 2 5 5 n g /k g * Y
8
Tissue Dose as Dose Metric for TCDD
• Based on Diliberto et al (1996 and 2001)
– Female B6C3F1 mice exposed oral gavage
• Single exposure 3 dose levels and 4 time points (7-35 days)
• Repeated dose study 5d/wk for up to 17 weeks
– dose response and time course study of disposition and enzyme
induction
• Fit the Hill model to the hepatic enzyme data using either
administered dose, tissue concentration, or body burden as the
dose metric
Office of Research and Development
National Health & Environmental Effects Research Laboratory
9
Figure 1a.
Liver Concentration vs. Hepatic EROD
Figure 1 c.
Blood Concentration vs. Hepatic EROD
(Combined Acute and Subchronic Data)
(Combined Acute and Subchronic Data)
18000
16000
16000
14000
14000
12000
10000
8000
6000
4000
Figure 1 b.
Body Burden vs. Hepatic EROD
(Combined Acute and Subchronic Data)
18000
16000
14000
Hepatic EROD
(pmol/min/mg protein)
18000
Hepatic EROD
(pmol/min/mg protein)
Hepatic EROD
(pmol/min/mg protein)
Relationship between different dose metrics and response to TCDD
12000
10000
8000
6000
4000
12000
10000
8000
6000
4000
2000
2000
2000
0
0
0
0
10
20
30
40
50
60
70
Liver Concentration
(ng TCDD/g)
Office of Research and Development
National Health & Environmental Effects Research Laboratory
0.00
0.02
0.04
0.06
Blood Concentration
(ng TCDD/g)
0.08
0.10
0
1000
2000
3000
4000
5000
6000
Body Burden
(ng TCDD/kg body wt)
10
7000
Dose Metrics for Developmental Effects
(Hurst et al., 1998, 2000)
Figure 1. Percent decrease in ejaculated sperm
count versus estimated mean fetal TCDD
concentration on GD16.
ƒ Tissue concentrations provide good
prediction of effect.
Figure 2. Puberty delay in males versus
estimated mean fetal concentration on GD16.
ƒ Tissue concentrations provide good
prediction of effect.
Office of Research and Development
National Health & Environmental Effects Research Laboratory
11
Benchmark Dose Approach
• Statistically-derived dose that gives a prespecified
increase in response or “risk”
• All dose-response data used
• Proposed as an alternative to the use of No-ObservedAdverse-Effect-Level (NOAEL) for estimating a point of
departure.
Office of Research and Development
National Health & Environmental Effects Research Laboratory
12
Hill Model
νdn
kn + dn
Response
R(d) = b +
R(∞)
Parameters
R(d) = Expected response
d = dose
ν
ED50
b = background response
k = ED50
n = Hill coefficient
ED01
R(0)
Dose
Office of Research and Development
National Health & Environmental Effects Research Laboratory
13
Point of Departure
• The point on a dose-response curve established from
experimental data, e.g., the benchmark dose,
generally corresponding to an estimated low effect
level ( e.g.,1% to 10% incidence of an effect ).
• Depending on the mode of action and available data,
some form of extrapolation below the POD may be
employed for low-dose risk assessment
• POD may be divided by a series of uncertainty factors
to arrive at a reference dose.
Office of Research and Development
National Health & Environmental Effects Research Laboratory
14
Response
• Quantal
– Incidence of disease or toxicity
– Can be described as a percent (ED50)
• Continuous
– Clinical, biochemical, physiological
– Can be described as a percent (ED50) but it is percent of maximum
• Bench Mark Response
– Some defined response related to adversity
– One standard deviation from control
– Percent of maximum response
Office of Research and Development
National Health & Environmental Effects Research Laboratory
15
Limitations and assumptions
• Any model is an a optimized approximation of the true dose
response from the available data.
• Dose # and spacing can have a big impact on the parameter
estimates.
– Choice of metric can define the spacing
• Ill defined maxima can lead to highly variable ED estimates
– E.g For incidence data Max =100%, is this true?
• Background response impacts ED estimate.
– Zero dose is not always zero exposure
Office of Research and Development
National Health & Environmental Effects Research Laboratory
16
Criteria for Study Inclusion in Modeling Analysis
ƒ 3 dose groups and a control
ƒ Data presented in a table
ƒ In some cases we contacted author for the data
> 300 Endpoints from 36 published manuscripts
– Single-dose (administration)
– Multiple repeated dose (administration)
• Response Categories
– Biochemical
– Hepatic
– Immune
– Retinol
– Thyroid
– Tissue
– Toxicity
Office of Research and Development
National Health & Environmental Effects Research Laboratory
17
Summary of Included Data
• 234 Responses with good model fits
– 43% Multiple-dose studies
– 32% Single-dose adult studies
– 25% Single-dose developmental studies
• Species comparison
– Rat (52%)
– Mouse (47%)
– Hamster (<1%)
Office of Research and Development
National Health & Environmental Effects Research Laboratory
18
Effects of Dioxins
Molecular Changes
Biochemical
Alterations
Altered
Metabolism
Cellular Effects
Altered
Proliferation/
Differentiation
Tissue/Organ
Effects
Altered
Homeostasis
Toxicity
Office of Research and Development
National Health & Environmental Effects Research Laboratory
19
Body Burden Values at the ED01
107
Body Burden ED01 (ng/kg)
106
105
104
103
102
101
1
Human body burden
10-1
10-2
10-3
Biochemical Hepatic
Immune
Retinol
Thyroid
Tissue
Assuming steady-state conditions, body burden calculated:
ED01 (ng/kg body burden) = ED01 (ng/kg/day) * half-life/ln(2) * f
Office of Research and Development
National Health & Environmental Effects Research Laboratory
20
EDs01 for Single-Dose Studies
Adult
109
Developmental
108
ED01 (ng/kg)
107
106
105
104
103
102
101
1
Office of Research and Development
National Health & Environmental Effects Research Laboratory
Toxicity
Tissue
Immune
Hepatic
Biochemical
Toxicity
Tissue
Immune
Hepatic
Biochemical
10-1
21
Shape of Dose Response
• 40% of all responses examined were linear: Hill coefficient < 1.5
• Study Type Comparison
– Multiple-dose: 43% linear
– Single-Adult: 43%
– Single-Developmental: 30%
• Category Comparison
– Retinol: 100%
– Thyroid: 50%
– Immune, Tissue, Biochemical: 40-45%
– Hepatic: 27%
– Toxicity: 20%
Office of Research and Development
National Health & Environmental Effects Research Laboratory
22
Conclusions
• EDs01/10 highly variable within and across response categories
• In general, half of the TCDD-induced responses support linearity and half
non-linearity
Office of Research and Development
National Health & Environmental Effects Research Laboratory
23
Mechanistic Modeling
• Physiologically-Based Pharmacokinetic (PBPK)
models
• Biologically-Based Dose Response (BBDR) models.
Office of Research and Development
National Health & Environmental Effects Research Laboratory
24
aerosol (broadcast, fogger)
Lung
Venous side
Arterial side
Generic PBPK
model structure
Brain
metabolism
metabolism
Viscera
Fat
iv dosing
dermal
Skin
Muscle
Dietary,
to mouth,
dietary, Hand
hand to
Mouth,
Object
to Mouth
object to
mouth
Liver
Volume of dist'n
Excretion
metabolism
Stomach
Office of Research and Development
National Health & Environmental Effects Research Laboratory
Gut
parent, metabolites in feces
25
Pharmacokinetics of dioxins
• Metabolism
– Limited
• Lipid partitioning
– Body fat mass: important at low exposures
• CYP1a2 sequestration
– Observed in animals and humans
– More important at high exposures
– CYP1A2 may play a role in metabolism
Increasing evidence that elimination is dose
dependent
Office of Research and Development
National Health & Environmental Effects Research Laboratory
26
Comparison of initial blood concentrations determined by first
order elimination or by PBPK model in 10 Ranch Hand Veterans
Groups
Low
High
Cblood in
1982
measured
[ pg/g lipid
adjusted]
12.7
16.7
23.5
24.6
25.0
33.7
43.8
115.5
182.3
209.7
Office of Research and Development
National Health & Environmental Effects Research Laboratory
Cblood at the time
discharge from
Vietnam estimated
using constant T½
[pg/g lipid adjusted]
53
44
72
112
83
103
123
381
602
640
Cblood at the time
discharge from
Vietnam estimated
using a PBPK model
[pg/g lipid adjusted]
138
166
277
587
168
492
197
6622
40376
35412
27
Cancer Modeling
• Empirical
– Ott and Zober (1996) (Hamburg cohort)
– Flesch-Janys et al (1998)
– Steenland et al (2001) (NOISH cohort)
• Mechanistic
– Single initiated phenotype
– Two initiated phenotype
Office of Research and Development
National Health & Environmental Effects Research Laboratory
28
Summary of Empirical Models
• Epidemiological data not sufficient to mandate a particular model shape, including linear
• These analyses assumed a first order elimination rate
of approximately 7 years in order to back calculate to
the initial exposures
• If the pharmacokinetics of TCDD are not first order, these assumptions are not valid and could lead to misleading results on the shape of the dose response relationships. (Cheng et al, 2006)
Office of Research and Development
National Health & Environmental Effects Research Laboratory
29
Kim et al (2003) Study Design
• Female Sprague Dawley rats
• Initiated with necrogenic dose of DEN
• Dioxin exposure started 2 weeks after DEN initiation
• Study terminated after 15 weeks
Office of Research and Development
National Health & Environmental Effects Research Laboratory
30
Exposure paradigm from Kim et al 2003
Office of Research and Development
National Health & Environmental Effects Research Laboratory
31
Office of Research and Development
National Health & Environmental Effects Research Laboratory
32
NTP TEF study
• 2-yr bioassay on TCDD, 4-PeCDF, PCB126,
• TEF method predicted liver tumors in SD rats for a
mixture of TCDD, 4-PeCDF and PCB126 (Walker et al
2005)
• For TCDD all tumors had non-linear dose response
relationships
Office of Research and Development
National Health & Environmental Effects Research Laboratory
33
Mechanistic Modeling
• Single initiated Cell Model (Portier et al 1996; Moolgavcar et al 1996)
– Assumes a single initiated phenotype
– Predicts low dose linearity for cancer risk
– ED01 approximately 3 ng/kg
– Rat liver tumor model
– Homogenous liver model (well mixed)
• Two initiated cell phenotype models (Conolly and Andersen 1997)
– Spatial model
– Rat liver tumor model
Office of Research and Development
National Health & Environmental Effects Research Laboratory
34
Biological Basis for dose response relationships
Office of Research and Development
National Health & Environmental Effects Research Laboratory
Ferrel and Machleder (1998)
35
Office of Research and Development
National Health & Environmental Effects Research Laboratory
Tritscher et al., 1992
36
All or none Phenomena for CYP induction
• TCDD
–Tritscher et al
–Mills et al
• Phenobarbital
–Jurtle et al.
–others
• What about other biochemical changes?
• What are implications for toxicity and carcinogenicity?
Office of Research and Development
National Health & Environmental Effects Research Laboratory
37
What can mechanistic modeling provide?
• Focus on one type of tumor – rat liver tumors
• In both humans and rodents we see multiple tumor
types
–Rats
• Liver, lung, oral mucosa, others –Humans
• All tumors
• Can mechanistic modeling inform on all of these tumor
types?
Office of Research and Development
National Health & Environmental Effects Research Laboratory
38
New Challenges since completion of latest version of the Dioxin reassessment
• Dose dependent pharmacokinetics
–Potential impact on exposure estimates in epidemiological studies.
–Michalek et al 2003; Aylward et al 2005
–Potential to impact cancer risk (Cheng et al 2006)
• AUC as a dose metric (Kim et al 2003)
–Suggests that AUC may not be the entire story.
Office of Research and Development
National Health & Environmental Effects Research Laboratory
39
ACKNOWLEDGMENTS
• National Research Council
– Claude Emond
• NIEHS
– Chris Portier
– Nigel Walker
– Fred Parham
• UNC-CH
– Amy Kim
Office of Research and Development
National Health & Environmental Effects Research Laboratory
•
USEPA
– Bill Farland
– Linda Birnbaum
– Matt Lorber
– Dwain Winters
– David Cleverly
– John Schaum
– Linda Tuxen
– Janet Diliberto
– David Ross
– Vicki Richardson
40